No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Applying support vector machines to predict building energy consumption in tropical region
Abstract
The methodology to predict building energy consumption is increasingly important for building energy baseline model development and measurement and verification protocol (MVP). This paper presents support vector machines (SVM), a new neural network algorithm, to forecast building energy consumption in the tropical region. The objective of this paper is to examine the feasibility and applicability of SVM in building load forecasting area. Four commercial buildings in Singapore are selected randomly as case studies. Weather data including monthly mean outdoor dry-bulb temperature (T0), relative humidity (RH) and global solar radiation (GSR) are taken as three input features. Mean monthly landlord utility bills are collected for developing and testing models. In addition, the performance of SVM with respect to two parameters, C and ɛ, was explored using stepwise searching method based on radial-basis function (RBF) kernel. Finally, all prediction results are found to have coefficients of variance (CV) less than 3% and percentage error (%error) within 4%.
... The optimal operation of a building necessitates a multifaceted strategy that minimizes energy consumption while maintaining a comfortable and healthy environment for occupants. The ability to predict electricity consumption is of paramount importance in this endeavor [5,7]. ...
... A plethora of prediction tools are currently employed to forecast electricity consumption in buildings. These tools encompass a spectrum of methodologies, each of which is designed to address a specific set of needs and to offer a unique set of advantages [1][2][3][4], e.g., machine learning algorithms [11,12], which include support vector machines (SVMs) [7,13,14], artificial neural networks (ANNs) [3,[15][16][17][18], decision trees [3], linear regression [6,15,19,20] and other statistical algorithms [8,12,[21][22][23]. ...
... The predominant focus of existing research in this field lies in either building energy consumption [1,3,[5][6][7]13,14] or the prediction and utilization of daylight availability [10][11][12]16,[20][21][22]. Nevertheless, a reliance on daylight alone is often inadequate. ...
This article presents a comparative analysis of two prominent machine learning techniques for predicting electricity consumption in workplace lighting systems: polynomial regression analysis and artificial neural networks. The primary objective is to assess their suitability and applicability for developing an accurate predictive model. After a brief overview of the current state of energy-saving techniques, the article examines several established models for predicting energy consumption in buildings and systems. These models include artificial neural networks, regression analysis and support vector machines. It then focuses on a practical comparison between polynomial regression analysis and an artificial neural network-based model. The article then looks at the data preparation process, outlining how the data is used within each model to establish appropriate prediction functions. Finally, it describes the methods used to evaluate the accuracy of the developed prediction functions. These functions allow the prediction of lighting consumption based on external lighting intensity. The article evaluates the accuracy of the developed prediction functions using the root mean square error, correlation coefficient and coefficient of determination values. The article compares these values obtained for both models, allowing a conclusive assessment of which model provides superior accuracy in predicting lighting consumption based on external lighting intensity.
... • Electric consumption [36], [37] • Electric consumption [38], [39] • Cooling load [40] • Hourly [37] • Monthly [36] • Daily and Halfhourly [38] • Daily [40] • Monthly [39] • Calendar/historical data/Temp • Sensor/Temp/holidays indicator • Meters • Measure data • Historical data 2 years 1 year 6 months 3 years ...
... • Electric consumption [36], [37] • Electric consumption [38], [39] • Cooling load [40] • Hourly [37] • Monthly [36] • Daily and Halfhourly [38] • Daily [40] • Monthly [39] • Calendar/historical data/Temp • Sensor/Temp/holidays indicator • Meters • Measure data • Historical data 2 years 1 year 6 months 3 years ...
... The information from the management system corresponds to one year of consumption. Dong [36] forecasts electricity consumption using an SVR algorithm for a set of commercial buildings. The input variables were the monthly electric service billing and weather data for four years [37], [38], [39], [40], [41], [42], [43]. ...
Analyzing the energy consumption behavior in buildings is essential for implementing energy-saving and efficient energy use measures without losing attention to the comfort inside the buildings. In this study, a statistical analysis and time series forecast of the energy situation of a group of buildings in a university academic unit in Mexico City was conducted. Seasonal Autoregressive Integrated Moving Average (SARIMA) models were used for the forecast with electrical energy consumption data from 55 months. Training and test partitions were created with these data to generate two SARIMA models. The results showed a strong dependence on the school cycle of electricity consumption, in addition to a shift in the cycle in the first year of the study. The mean absolute percentage error (MAPE) for the training partitions created shows that the best fit is provided by the SARIMA (3,1,1) (1,0,0)12 model for the 48-month separation. In comparison, the SARIMA (2,1,2) (1,0,0)12 model does so for the 43-month test partition. The confidence intervals for the 7- and 12-month forecast are less wide for the SARIMA (3,1,1) (1,0,0)12 model than for the SARIMA (2,1,2) (1,0,0)12 model. Statistical analysis and time series modeling allows a better understanding of the building stock's energy performance and strengthens the energy audit to design or implement energy saving or efficient energy use measures.
... The use of AI in this context signals a transformation, indicating a new era of elevated energy efficiency, sustainability, and environmental stewardship in residential building design and operation. Various AI and ML models have been applied in computational analysis and prediction of building efficiency in terms of HL and Cl, such as artificial neural network (ANN), support vector regression (SVR), and linear and polynomial least square regression (PLS) [8,18,19]. For instance, Zhang and Haghighat [20] developed a regression approach to predict monthly heating demand for single-family homes in temperate climates. ...
... The authors are thankful to the Deanship of Scientific Research at Najran University for funding this work under the Research Priorities and Najran Research funding program grant code (NU/NRP/SERC/12/2). 18 International Journal of Energy Research ...
Buildings play a critical role in energy consumption, representing one of the primary consumers of power. Heating load (HL) and cooling load (CL) are essential for determining the energy efficiency of buildings. Several research projects attempt to address the critical challenge of enhancing energy efficiency in residential buildings, focusing on accurately estimating HL and CL using solutions that implement statistical prediction or typical building control management. This study, however, looked into advanced machine learning (ML) models for sustainable building design based on harnessing the potential of artificial intelligence and explainable AI (AIX) technologies. The proposed model was trained and tested using a dataset of 768 buildings based on feature engineering methods with various ML algorithms (including cutting-edge emotional neural learning (ENN), nonparametric kernel-based probabilistic models known as Gaussian process regression (GPR), and boosted tree (BT) algorithm). In addition, the output of the model was fed to standard building energy performance software (Ecotect) that utilizes the dataset from twelve different building shapes to perform various building energy efficiency analyses. The overall performance of the proposed model was measured using different performance metrics, including MAPE, MAE, RMSE, and PCC to measure the performance of HL- and CL-based building energy efficiency. The performance evaluation results indicate that the M3 variants, especially GPR-M3, consistently outperformed their counterparts across heating and cooling scenarios. The three models indicated reliability for modeling HL and CL. However, for HL, the GPR-M3 model emerged as the best model, outperforming GPR-M1 by 9.2% and GPR-M2 by 3.9%. Similarly, GPR-M3 is superior to CL, with the highest PCC at 0.9858, marking an 8.1% and 1.9% improvement over GPR-M1 and GPR-M2, respectively.
... The data-driven approach, learning from historical data by employing machine learning models without the use of the on-site physical information of buildings, has been more commonly applied in recent studies because the complex building information that is required in the physical modeling approach can lead to high computational cost and low running speed [5]. Recent review studies [5,6] have shown that artificial neural networks (ANNs) [7,8] support vector machine (SVM) [9,10], statistical regression [11,12], and decision tree-based models (DT) [13] are the most popular data-driven methods for building energy baseline modeling. However, the literature lacks a comprehensive comparative analysis across all major types of buildings. ...
... Comparisons with the conventional regression model [16] and Energy Plus [17] showed that neural networks generated more accurate energy predictions, especially when energy consumption shows high fluctuation. Dong et al. [9] and Li et al. [10] validated the effectiveness of the SVM models in predicting energy consumption in commercial buildings and showed a better performance than the conventional back-propagation neural networks. ...
Building energy baseline models, particularly machine learning-based models, are a core aspect in the evaluation of building energy performance to identify inefficient energy consumption behavior. In smart city design, energy planners and decision makers require comprehensive information on energy consumption across diverse building types as well as comparisons between different types of buildings. However, there is no comprehensive study of baseline modeling across the main building types to help identify factors that influence the performance of different machine learning algorithms for baseline modeling. Therefore, the goal of this paper is to review and analyze energy consumption behavior and evaluate the prediction performance and interpretability of machine learning-based baseline modeling techniques across major building types. The results have shown that the Extreme Gradient Boosting Machine (XGBoost) model is the most accurate baseline modeling method for all building types. Time-related factors, especially the week of the year and the day of the week, have the most impact on energy consumption across all building types. This study is presented as a useful resource for smart city energy managers to help in choosing and setting up appropriate methodologies for better operational effectiveness and efficiencies when designing and planning smart energy systems.
... Dong et al. [63] used SVMs to predict building energy consumption in a tropical region, considering four case studies in Singapore without neglecting the dynamic parameters such as external temperature and humidity. Similarly, Ahmad et al. [64] applied SVMs to predict data derived from a solar thermal collector. ...
... As already anticipated in Section 2.1.3, in order to predict building energy consumption, Dong et al. [63] used SVMs. The influence of outdoor climatic conditions was considered too, e.g., relative humidity, solar radiation and monthly mean external temperature. ...
Given the climate change in recent decades and the ever-increasing energy consumption in the building sector, research is widely focused on the green revolution and ecological transition of buildings. In this regard, artificial intelligence can be a precious tool to simulate and optimize building energy performance, as shown by a plethora of recent studies. Accordingly, this paper provides a review of more than 70 articles from recent years, i.e., mostly from 2018 to 2023, about the applications of machine/deep learning (ML/DL) in forecasting the energy performance of buildings and their simulation/control/optimization. This review was conducted using the SCOPUS database with the keywords “buildings”, “energy”, “machine learning” and “deep learning” and by selecting recent papers addressing the following applications: energy design/retrofit optimization, prediction, control/management of heating/cooling systems and of renewable source systems, and/or fault detection. Notably, this paper discusses the main differences between ML and DL techniques, showing examples of their use in building energy simulation/control/optimization. The main aim is to group the most frequent ML/DL techniques used in the field of building energy performance, highlighting the potentiality and limitations of each one, both fundamental aspects for future studies. The ML approaches considered are decision trees/random forest, naive Bayes, support vector machines, the Kriging method and artificial neural networks. The DL techniques investigated are convolutional and recursive neural networks, long short-term memory and gated recurrent units. Firstly, various ML/DL techniques are explained and divided based on their methodology. Secondly, grouping by the aforementioned applications occurs. It emerges that ML is mostly used in energy efficiency issues while DL in the management of renewable source systems.
... In order to circumvent the computational burden of physics-based models and enable a wider exploration of the problem space, some works concentrated on using surrogate modeling or other machine learning techniques. In [15,16], Multiple Linear Regression (MLR) [17] was used to predict the energy consumption based on weather data (temperature, humidity, radiation), and information from weather stations and commercial buildings in Singapore to train the model. An Artificial Neural Network (ANN) was used to predict energy consumption for commercial buildings in Hawaii, United States, using as input climatic variables [18] and building properties [19], reaching satisfactory correlation levels. ...
... In this example, the uncertainty from the physics-based model (ε CEA ) is not considered (µ = 0, σ = 0). 16) Therefore, the probabilities of the EUI are the result of applying the observed properties into the reduced order SM. The results for each scenario can be analysed via probability density functions (PDF) and cumulative distribution functions (CDF), as presented in Figure 8. ...
Building energy demand impacts a myriad of interconnected economic, societal, and environmental aspects. As a result, Buildings Energy Models (BEM) play an important role in the process of urban design and planning. While previous studies have investigated the effects of building interventions on energy efficiency, their applicability may be limited due to the BEM’s high computational complexity. This limits their ability to systematically study important aspects of energy demand on a large scale. The development of Machine Learning Models (MLM) allows to design the required detailed analysis and solutions, while reducing the computational burden, making MLM attractive for urban designers. The capability of MLM to generalize well for multiple contexts (in our case, multiple buildings) is a crucial contributor to their applicability. However, the validation process in a wider context is often overlooked, therefore its generalization capabilities are not quantified. In this paper, we present a framework to train and validate a surrogate model derived from a physics-based BEM. Our method employs a Multiple Linear Regression model to predict Energy Use Intensity (EUI) for office buildings in Singapore using 36 input parameters (covariates), based on a training dataset of 23,000 samples. Model validation is performed by comparing the results of the Surrogate Model (SM) to a widely used BEM for a sample of 120 buildings. Our results indicate that the SM has an accuracy of NRMSE of 13%, NMBE of −3.56%, and R2 of 0.92, which suggests it can effectively and accurately predict building EUI. We also conduct a sensitivity analysis, which indicates that the parameters associated with internal loads and internal space usage are the most influential. Additionally, we present a reduced order model trained with only the 11 most influential parameters, which exhibits negligible loss in accuracy compared to the full SM while providing reduced complexity. Finally, we demonstrate an application of our SM to evaluate energy efficiency under uncertainty scenarios. The analytically derived results indicate a potential reduction of EUI of offices in Singapore from 227kWh/m2 to 99kWh/m2 by altering the building parameters that were identified as most influential.
... With the recent rise of artificial intelligence and machine learning, more work is being performed to integrate machine learning techniques into the field. This can be identified in numerous studies [2][3][4][5][6], giving researchers the opportunity to utilize machine learning tools to study the effect of numerous building parameters on energy-based outputs, making the procedure more efficient if a database of similar structure is available. ...
... When reviewing the overall appearance of machine learning in energy efficiency, it can be seen that various models such as polynomial regression [19], support vector machines (SVM) [4,20], artificial neural networks (ANNs) [21,22], and decision trees [5,6] have been utilized to predict specific variables within the energy efficiency field. Machine learning tools have also been explicitly used in predicting appliance energy use in other studies. ...
Due to the transition toward the Internet of Everything (IOE), the prediction of energy consumed by household appliances has become a progressively more difficult topic to model. Even with advancements in data analytics and machine learning, several challenges remain to be addressed. Therefore, providing highly accurate and optimized models has become the primary research goal of many studies. This paper analyzes appliance energy consumption through a variety of machine learning-based strategies. Utilizing data recorded from a single-family home, input variables comprised internal temperatures and humidities, lighting consumption, and outdoor conditions including wind speed, visibility, and pressure. Various models were trained and evaluated: (a) multiple linear regression, (b) support vector regression, (c) random forest, (d) gradient boosting, (e) xgboost, and (f) the extra trees regressor. Both feature engineering and hyperparameter tuning methodologies were applied to not only extend existing features but also create new ones that provided improved model performance across all metrics: root mean square error (RMSE), coefficient of determination (R2), mean absolute error (MAE), and mean absolute percentage error (MAPE). The best model (extra trees) was able to explain 99% of the variance in the training set and 66% in the testing set when using all the predictors. The results were compared with those obtained using a similar methodology. The objective of performing these actions was to show a unique perspective in simulating building performance through data-driven models, identifying how to maximize predictive performance through the use of machine learning-based strategies, as well as understanding the potential benefits of utilizing different models.
... The SVR method approximates the function by the usage of the following expression [40,41]: ...
... Thus, by substituting Eq. (35), Eq. (39), and Eq. (40) in Eq. (34), the position of the grasshoppers can be expressed as [46]: ...
Solar Chimney Power Plant (SCPP) is a renewable energy system that indirectly converts solar energy to electricity. However, the efficiency of SCPP is not sufficient for practical applications. Integrating SCPPs with Photovoltaic-Thermal systems (PVT) could enhance their performance to levels acceptable for industrial adoption. This study investigates the combined SCPP-PVT performance for the weather conditions of Austin (Texas), San Diego (California), and Phoenix (Arizona), all on a similar latitude. Various configurations of this combined system are numerically simulated, and their efficiencies are compared with a conventional SCPP, SCPPPhotovoltaic (PV), and stand-alone PV modules. Moreover, to predict and optimize the performance of these systems, the Support Vector Regression with Linear (LSVR), Polynomial (PSVR), Gaussian (GSVR), and Hybrid (HSVR) kernels are implemented. In order to optimize the hyperparameters of the Machine Learning (ML) models, the Grey Wolf Optimizer (GWO) is implemented. Also, the optimum performance of the SCPP-PVT system is obtained using the Multi-objective Grasshopper Optimization Algorithm (MOGOA). The results show that the HSVR ML model has the highest accuracy, followed by PSVR, GSVR, and LSVR models. It is shown that the SCPP-PVT system outperforms both SCPP-PV and stand-alone PV modules, respectively. Finally, the SCPP-PVT is shown to outperform the PV modulus by up to 4.8%.
... gorized as a new neural network algorithm for forecasting (Dong, Cao, and Lee 2005). Besides that, since it can be used to solve nonlinear regression estimation problems, SVM can be used to forecast time series. ...
Ceiling fans have been promoted as an alternative cooling technique to save energy, while their adoptions in the residential sector have yet to be investigated. This study analysed the adoption of ceiling fans in residential buildings and compared their energy-saving potential with that of airconditioning in Building Sustainability Index (BASIX) certificated single (n = 268,558) in New South Wales, Australia, from 2011 to 2019. Four climate zones were classified according to heating and cooling hours via k-means, based on which predictive models for the relationship between cooling technologies and energy-saving levels were established via machine learning. Dwellers in the hot zone of New South Wales would be more likely to adopt ceiling fans. In the living rooms, dwellers could adopt ceiling fans alone, while in bedrooms, dwellers could adopt airconditioning in addition to ceiling fans. This study provides empirical evidence on the adoption of ceiling fans in green buildings and helps to map out the low carbon solutions using alternative cooling in the residential sector. ARTICLE HISTORY
... Support Vector Machines (SVM). In 2005, Dong, Cao, and Lee (2005) examined the applicability of Support Vector Machines (SVM) in building energy consumption prediction. In this study, the input variables consisted of three types of monthly weather data, dry-bulb temperature, relative humidity, and global solar radiation. ...
Achieving energy efficiency, zero emissions, resiliency, and healthfulness are important subjects facing the current field of buildings. Building performance simulation is the means established and accredited to quantify building performance and thus enables the communication of building energy efficiency and carbon emissions information. However, traditional building performance physics-based simulation presents significant challenges and shortcomings, such as being complex, time-consuming, and divergent when compared to actual performance. There remains a need for ease of use, immediate simulation, and accurate building performance prediction approaches. Emerging statistical and machine learning techniques open the possibility of developing a novel prediction model with ease of use, immediate simulation, and accurate features. Investigating the application of statistical and machine learning techniques in building performance prediction has been an attractive research direction. However, the published knowledge on applying statistical and machine learning techniques in building performance prediction remains inadequate and insufficient in terms of scalability and universality. This paper provides an up-to-date review of the application of statistical and machine learning techniques in building performance prediction, intending to recommend the latest research status and enlighten future research points. A comprehensive discussion on the impetus and strengths of applying statistics and machine learning in building performance is highlighted. In contrast, the limitations of existing applications and recommendations for future research are pronounced. Distinct from similar reviews that cover a broader application range, this paper focuses on the prediction application of whole building performance.
... Numerous studies have employed machine learning for building energy consumption predictions [9,10]. Furthermore, artificial neural networks (ANN) [11], support vector machines (SVM) [12], and extreme learning machines [13] are also advanced data-driven methods for building energy prediction and have already been widely applied in this field. ...
With the advancement of information technology, energy consumption prediction models are widely used for various types of buildings (office, residential, and commercial buildings) as guidance during the design and management stages. This article will establish an efficient building energy consumption prediction model for hotel buildings. To achieve this, we collected 78 architectural drawings of high-rise hotel buildings to establish 6 kinds of typical energy consumption models in 2 standard floor layouts and 3 public area levels. Then, on this basis, we used the total energy consumption calculated by EnergyPlus as an indicator to conduct sensitivity analysis on geometric feature parameters, internal heat source parameters, and thermal parameters, respectively. Finally, we generated a building database with 5000 samples through the R programming language to calculate and verify the energy consumption. As a result, it was proved that the energy consumption of hotel buildings can be predicted accurately, and that quadratic polynomial regression, with the best accuracy and stability, is the most suitable optimization model for hotel energy consumption prediction in Guangzhou. These conclusions provide a good theoretical basis for the analysis, prediction, and optimization of energy consumption in high-rise hotel buildings in the future.
... rather than only reducing the training error (Dong et al. 2005). Support vector regression (SVR) is the name of the SVM's regression variant. ...
Ghana has witnessed major flood and drought events as a result of unpredictable climate conditions, affecting numerous sectors of the economy such as agriculture, mining, hydroelectric power, and so on. Forecasting rainfall on monthly or seasonal timescales will be critical in averting or mitigating the consequences of these natural disasters, particularly given the present danger posed by climate change. The suitability of long short-term memory (LSTM) recurrent neural networks for monthly rainfall forecasting in Ghana has been investigated in this study. Both univariate and multivariate LSTM models are examined. The numerical results showed that the proposed models performed well in forecasting monthly rainfall in the selected locations across the country. There was no significant difference in performance between the two LSTM models. The values of coefficient of correlation (R), Nash-Sutcliffe efficiency coefficient (NS), root mean squared error (RMSE), and mean absolute error (MAE) ranged from 0.604 to 0.894, 0.352 to 0.679, 50.21 to 75.89, and 36.53 to 44.56, respectively, for the univariate LSTM model, and 0.693 to 0.890, 0.357 to 0.655, 42.10 to 77.04, and 33.07 to 58.60, respectively, for the multivariate LSTM model. The LSTM models’ performance was compared to two standard machine learning models: support vector machine and random forest. Overall, the proposed LSTM recurrent networks outperformed the standard SVR and RF forecasting models. However, all the models (LSTM, SVR, and RF) considered in this study can be used to forecast monthly rainfall across the country with reasonable accuracy.
... To overcome these limitations, several researchers have employed a supervised learning (SL) approach, including random forest (RF) [12], support vector regression (SVR) [13,14] and extreme learning [15], artificial neural network [16], and deep neural network [17]. These methods were trained by searching for best parameters that describe the hidden relationships between inputs and outputs and yield the lowest errors for the test instances. ...
As the scale of electricity consumption grows, the peak electricity consumption prediction of campus buildings is essential for effective building energy system management. The selection of an appropriate model is of paramount importance to accurately predict peak electricity consumption of campus buildings due to the substantial variations in electricity consumption trends and characteristics among campus buildings. In this paper, we proposed eight deep recurrent neural networks and compared their performance in predicting peak electricity consumption for each campus building to select the best model. Furthermore, we applied an attention approach capable of capturing long sequence patterns and controlling the importance level of input states. The test cases involve three campus buildings in Incheon City, South Korea: an office building, a nature science building, and a general education building, each with different scales and trends of electricity consumption. The experiment results demonstrate the importance of accurate model selection to enhance building energy efficiency, as no single model’s performance dominates across all buildings. Moreover, we observe that the attention approach effectively improves the prediction performance of peak electricity consumption.
... Even as data-driven BPSs evolve, challenges persist, from computational intensity to potential overestimations of the energy savings [7], from underlying biases to unclear data processing. However, recent research [8,9] shines a spotlight on the potential of data-driven methods, with techniques like Artificial Neural Networks (ANNs) [10,11], Support Vector Machine (SVM) [12,13], and Decision Tree-based models (DT) [14] gaining traction in building energy modelling [15]. ...
Cities and buildings represent the core of human life, the nexus of economic activity, culture, and growth. Although cities cover less than 10% of the global land area, they are notorious for their substantial energy consumption and consequential carbon dioxide (CO2) emissions. These emissions significantly contribute to reducing the carbon budget available to mitigate the adverse impacts of climate change. In this context, the designers’ role is crucial to the technical and social response to climate change, and providing a new generation of tools and instruments is paramount to guide their decisions towards sustainable buildings and cities. In this regard, data-informed digital tools are a viable solution. These tools efficiently utilise available resources to estimate the energy consumption in buildings, thereby facilitating the formulation of effective urban policies and design optimisation. Furthermore, these data-driven digital tools enhance the application of algorithms across the building industry, empowering designers to make informed decisions, particularly in the early stages of design. This paper presents a comprehensive literature review on artificial intelligence-based tools that support performance-driven design. An exhaustive keyword-driven exploration across diverse bibliographic databases yielded a consolidated dataset used for automated analysis for discerning the prevalent themes, correlations, and structural nuances within the body of literature. The primary findings indicate an increasing emphasis on master plans and neighbourhood-scale simulations. However, it is observed that there is a lack of a streamlined framework integrating these data-driven tools into the design process.
... In the SVR model, the function is approximated as the following expression [37,38]: ...
Integrating conventional Heat Sinks with Phase Change Materials (HS-PCM) is a novel strategy for the thermal management of electronic devices. Although various studies attempted to investigate the impact of PCM on the operating temperature of electronic devices, the influence of PCM properties on the performance of the HS-PCM system remained unknown. Therefore, in the present study, an accurate three-dimensional HS-PCM system is designed and simulated to evaluate the effect of PCM properties, including thermal conductivity, enthalpy, specific heat capacity, and liquidus temperature, on the operating temperature of this system. In this order, the HS-PCM system is simulated at multiple operating conditions during both the working and cooling phases. Notably, an experimental setup is fabricated to compare the numerical outputs against the experimental data. Additionally, the Support Vector Regression with Hybrid (HSVR) kernels is used to predict the performance of the HS-PCM system. The hyperparameters of this machine learning model are optimized by implementing Grey Wolf Optimizer (GWO). It is found that raising the liquidus temperature of the PCM can have a minor impact on the working duration of the chipset while it dramatically impacts the cooling process of the system. Also, increasing the enthalpy and heat capacity of the PCM has a favorable effect on the working time of the chipset but extends the cooling process. Thermal conductivity is the only parameter which its enhancement has a positive impact on both the working and cooling duration of the chipset. Raising the thermal conductivity of the PCM from 0.05 W/(m⋅K) to 0.8 W/(m⋅K) improves the working duration of the chipset from 22 min to 24 min. Based on the outcomes, the most effective factor on the working time of the chipset is the enthalpy, followed by heat capacity, thermal conductivity, and the liquidus point of the PCM. According to the findings, the GWOHSVR model can accurately forecast the outcomes of the HS-PCM system at the both working and cooling phases. In the testing process of the model, the values of R2 for the working and cooling phases are calculated to be 0.99174 and 0.99775, respectively.
... As a corollary, these avant-garde, data-driven energy models possess the capability to quantify the ramifications of myriad energy determinants, encompassing building characteristics, meteorological conditions, spatial patterns, and utilization typologies, achieving elevated precision benchmarks (Zhao & Magoulès, 2012). Multiple linear regression models (Asadi, Amiri, & Mottahedi, 2014), support vector regression (Dong, Cao, & Lee, 2005), decision trees (Yu, Haghighat, Fung, & Yoshino, 2010), and artificial neural networks (Kim, Kim, & Srebric, 2020) are typical models that have been used in the past to estimate the energy consumption of buildings. The energy usage of multi-family residential structures has been modeled using support vector regressions, kernel-based algorithms that address nonlinear issues (Nutkiewicz et al., 2018). ...
Predicting solar radiation in cities using the Artificial Neural Network model (ANN) is a pioneering step in transforming future-oriented city planning using solar energy. This research harnesses vast datasets to forecast the average annual solar radiation, considering minimal urban information across various urban attributes, including coordinates (X, Y, Z), average height, inhabited and non-occupied areas, and the Azimuth angle. Our method employed parametric design and remote sensing to generate this dataset and then used the ANN model to make predictions and simulations. Urban attributes of 20 cities were examined, including Casablanca, Abu Dhabi, Cape Town, Dublin, Havana, Melbourne, Rome, Singapore, Nairobi, Mumbai, New York, Nagoya, Sao Paulo, Tehran, Madrid, Toronto, Antananarivo, Beijing, Lisbon, and Paris. This data-driven approach trains our ANN model to discern complex and nonlinear relationships between independent and dependent variables and thus enables our model to predict solar radiation in urban cities. Our data training results indicate that the output (the minimum solar radiation each year of the cities) can be predicted using the study input variables with a loss of 0.01, a mean squared error of 0.01, and an R2-squared value of 85%. Such predictions can refine urban designs of buildings, public spaces, and various urban infrastructures to optimize solar energy use, reducing environmental impacts and fossil fuel reliance, thus aiding climate change mitigation and sustainability. Our findings underscore the integral association between solar radiation and sustainable urban evolution, giving urban planners and researchers sustainable strategies for advancing energy efficiency and ecological equilibrium.
... Such energy sources have the potential to reduce carbon emissions by reducing environmental impacts compared to energy production based on fossil fuels [2]. At the same time, renewable energy significantly contributes to energy security because it relies on locally available resources and helps reduce energy imports [3]. In the future, renewable energy technologies are expected to develop rapidly, and their costs to decrease. ...
This study focuses on estimating the energy amount of the solar energy system installed on the house roof with the Adaboost Machine Learning (ML) algorithm. Root Mean Square Error (RMSE), Mean Squared Error (MSE), and Mean Absolute Error (MAE) values of the model were calculated as 0.005, 0.068, and 0.015, respectively. The R2 value of the algorithm was calculated as 0.995, indicating that the independent variables of the model have a very high ability to explain the dependent variable. These results show that this model can be used safely in critical applications such as solar energy production or consumption. This study shows that the Adaboost algorithm is a powerful model for solar energy prediction, and its predictions are incredibly close to actual values. In conclusion, this study is essential for homeowners and energy providers as accurate estimates of electricity consumption will help in more effective energy management and efficient use of resources. This work will lead to critical applications in energy management and effective energy use.
... Based on the feature importance analysis results, time of year emerges as the most significant factor, with an importance of 49.444 %. This feature serves as a proxy for weather conditions, which are widely recognized as influential factors in predicting energy consumption [23,74,83,[75][76][77][78][79][80][81][82]. The next most influential factor is the number of occupants, which has an importance of 5.633 % (Fig. 12). ...
... Even with limited observations from the training datasets, the approach can effectively find an excellent solution to nonlinear problems [48,72]. The structural risk minimization (SRM) concept, which SVM employs, seeks to minimize the upper and lower bounds of the generalization error, which is the sum of the training error and a confidence level [73,74]. This concept is distinct from the more common empirical risk minimization (ERM) method, which focuses only on reducing the total error generated during the training stage. ...
The Great Lakes are critical freshwater sources, supporting millions of people, agriculture, and ecosystems. However, climate change has worsened droughts, leading to significant economic and social consequences. Accurate multi-month drought forecasting is, therefore, essential for effective water management and mitigating these impacts. This study introduces the Multivariate Standardized Lake Water Level Index (MSWI), a modified drought index that utilizes water level data collected from 1920 to 2020. Four hybrid models are developed: Support Vector Regression with Beluga whale optimization (SVR-BWO), Random Forest with Beluga whale optimization (RF-BWO), Extreme Learning Machine with Beluga whale optimization (ELM-BWO), and Regularized ELM with Beluga whale optimization (RELM-BWO). The models forecast droughts up to six months ahead for Lake Superior and Lake Michigan-Huron. The best-performing model is then selected to forecast droughts for the remaining three lakes, which have not experienced severe droughts in the past 50 years. The results show that incorporating the BWO improves the accuracy of all classical models, particularly in forecasting drought turning and critical points. Among the hybrid models, the RELM-BWO model achieves the highest level of accuracy, surpassing both classical and hybrid models by a significant margin (7.21 to 76.74%). Furthermore, Monte-Carlo simulation is employed to analyze uncertainties and ensure the reliability of the forecasts. Accordingly, the RELM-BWO model reliably forecasts droughts for all lakes, with a lead time ranging from 2 to 6 months. The study’s findings offer valuable insights for policymakers, water managers, and other stakeholders to better prepare drought mitigation strategies.
... Following the method of Ihara et al. (2008), these five parameters were estimated (see Nakajima et al. 2022 for details). Other climatic factors (e.g., air humidity, solar radiation, and the heat index) have also been used to predict EC in similar models (Xu et al. 2020;Ihara et al. 2008;Hashimoto et al. 2019;Mirasgedis et al. 2006;Dong et al. 2005). The equation including humidity is as follows: ...
... The last ML model used to analyze and forecast Qatar's electricity consumption is the SVR method. This method is well-known among researchers in load forecasting, such as predicting the energy consumption of buildings (Dong, Cao, & Lee, 2005). The structure of SVR is based on the risk minimization assumption, as it aims to minimize the upper bound of the generalization error instead of finding out the empirical error. ...
... These algorithms are used for energy prediction because of their ability to handle complex non-linear data and large datasets [17,47]. For example, Shao et al. [48], Jain et al. [49], Zhao and Magoules [50], Lai et al. [51], Zhang et al. [52], Fan et al. [43], and Dong et al. [53] forecasted energy consumption using an SVM-based prediction model. Other studies, such as Ciulla et al. [54], Lei et al. [55], Yan et al. [56], and Singaravel et al. [57], used ANN-based prediction models. ...
... Revealing and establishing the inherent dependency relationship of AEU on potential drivers is very useful for energy control and management [4,5], building performance analysis through simulations [6,7], and energy consumption control [8]. Many methods have been proposed for AEU modelling and analysis, such as multiple linear regression [3], artificial neural networks [9,10], outlier detection [11], support vector machines [12], and model ensembles [13]. AEU is determined by many factors, e.g., local temperature in and outside the house, humidity in the building, time of the day, just mention a few [2]. ...
The modelling and analysis of appliance energy use (AEU) of residential buildings are important for energy consumption control, energy management and maintenance, building performance evaluation, and so on. Although some traditional machine learning methods have been applied to produce good prediction results, these models are usually not interpretable, in that they fail to explain how appliance factors make contributions to the variation of AEU individually and interactively. Explicitly knowing the role played by each of the appliance factors in explaining AEU, however, is very important for energy saving. Motivated by this observation, this study introduces an interpretable machine learning approach which is built upon the nonlinear autoregressive moving average with eXogenous inputs model. The advantage of the proposed model is that in comparison with other state-of-the-art machine learning methods, for example, feedforward neural network, recurrent neural network (e.g., gated recurrent unit), and long short-term memory network, the established model is not only able to produce more accurate energy use prediction, but more importantly, also fully transparent and physically interpretable, clearly and explicitly indicating which factors significantly affect the variation of AEU. The findings of this study provide meaningful insights for improving the AEU efficiency.
... Therefore, we need some powerful solutions to avoid significant energy demand estimations deviations. In the last decades, researchers proposed many valuable studies because energy consumption, distribution, and production have an essential role for countries [18][19][20][21][22]. Many algorithms based on different solution approaches were proposed in the literature, such as statistical applications [23][24][25][26][27][28][29], artificial neural networks-based methods (ANN) [30][31][32][33][34], and metaheuristic algorithms-based techniques [6, 12-14, 17, 22, 35-44] for generating the long-term energy demand estimations. ...
Energy consumption is getting rising gradually around the planet. Therefore, the importance of energy management has increased for all nations worldwide, and long-term energy demand estimation is becoming a vital problem for all countries. In this study, linear, quadratic and exponential models based six different Archimedes optimization algorithms (AOA) such as AOA-Linear, AOA-Quadratic, AOA-Exponential, IAOA-Linear, IAOA-Quadratic and IAOA-Exponential have been proposed to make some future projections of Turkey for the years (2021–2050). The previous studies in the literature were used the data set of Turkey, such as observed energy demand (OED), population, gross domestic product (GDP), export and import data for the years (1979–2005) or (1979–2011) obtained from the Turkish Statistical Institute (TUIK) and the Ministry of Energy and Natural Resources (MENR). However, in this study, a new data set is organized with the OED, population, GDP, export and import data of Turkey for the years (1997–2020) to make some long-term energy demand estimations of Turkey, and this dataset is used for the first time in this study. AOA-Linear, AOA-Quadratic and AOA-Exponential algorithms are based on linear, quadratic and exponential mathematical models and the basic AOA method. IAOA-Linear, IAOA-Quadratic and IAOA-Exponential algorithms are also based on linear, quadratic and exponential mathematical models and the improved AOA (For short, IAOA) proposed in this study. Once a sensitivity analysis is made for determining the effect of algorithmic parameters of AOA and IAOA, the proposed algorithms are realized for Turkey’s long-term energy demand estimation for the years (2021–2050) with three different future scenarios. According to the experimental results, the quadratic model-based proposed IAOA produces better or comparable performance on the problem dealt with in this study in terms of solution quality and robustness.
... In addition, SVMs are increasingly used in building energy simulations. Example areas include building energy consumption prediction (Dong et al., 2005;Jung et al., 2015), annual electric load forecast (Azadeh et al., 2008), estimating heating energy (Paudel et al., 2014), PV system performance (Ogliari et al., 2013), and wind power system (Chen et al., 2013). Studies show that SVMs can solve nonlinear problems with a small number of training data (Zhao et al., 2012). ...
... Furthermore, other researchers were interested in investigating other machine learning methods. The support vector machine (SVM) method was selected by Dong et al. [10] to predict the monthly energy consumed in four commercial buildings in Singapore. The effect of the model parameters was evaluated, and three weather parameters were considered as inputs: global solar radiation, outdoor temperature, and relative humidity. ...
Energy management is now essential in light of the current energy issues, particularly in the building industry, which accounts for a sizable amount of global energy use. Predicting energy consumption is of great interest in developing an effective energy management strategy. This study aims to prove the outperformance of machine learning models over SARIMA models in predicting heating energy usage in an administrative building in Chefchaouen City, Morocco. It also highlights the effectiveness of SARIMA models in predicting energy with limited data size in the training phase. The prediction is carried out using machine learning (artificial neural networks, bagging trees, boosting trees, and support vector machines) and statistical methods (14 SARIMA models). To build the models, external temperature, internal temperature, solar radiation, and the factor of time are selected as model inputs. Building energy simulation is conducted in the TRNSYS environment to generate a database for the training and validation of the models. The models' performances are compared based on three statistical indicators: normalized root mean square error (nRMSE), mean average error (MAE), and correlation coefficient (R). The results show that all studied models have good accuracy, with a correlation coefficient of 0.90 < R < 0.97. The artificial neural network outperforms all other models (R=0.97, nRMSE=12.60%, MAE= 0.19 kWh). Although machine learning methods, in general terms, seemingly outperform statistical methods, it is worth noting that SARIMA models reached good prediction accuracy without requiring too much data in the training phase. Doi: 10.28991/CEJ-2023-09-05-01 Full Text: PDF
... 11 The need for a faster and less computationally challenging method had given rise to an opportunity and increased acceptance in exploring the usage of machine learning approaches to analyse the effects of various building parameters with energy consumption. [13][14][15][16] Numerous studies had thus taken the opportunity to compare the model performances between the various machine learning models in order to generate greater understanding on their predictive natures. Candanedo et al. 17 compared the performance of Random Forest (RF) with Multiple Linear Regression (MLR), Support Vector Regression (SVR) and Gradient Boost Machines (GBM) in the prediction of short-term electricity consumption in Belgium houses. ...
Ensuring effective forecasting of buildings' energy consumption is crucial in establishing a greater understanding and improvement of buildings' energy efficiency. In Singapore, domestic electricity usage in public residential buildings takes up a significant portion of the country's annual energy consumption. Having effective forecasting approaches is thus important in supporting relevant strategies and policy making. In this research, we proposed a hybrid approach that was based on a combination of building characteristics and urban landscape variables to predict residential housing electricity usage in Singapore. XGboost was also incorporated inside the hybrid approach as the preferred machine learning approach for energy consumption predictions. To demonstrate our proposed approach's predictive strength, the performance of our proposed hybrid machine learning approach was compared with two other models, Geographically Weighted Regression (GWR) model and the Random Forest (RF) model. Results showed that our proposed hybrid model had outperformed these abovementioned approaches with higher accuracy (r ² value of 0.9). The proposed approach had thus been effective in forecasting electricity consumption for public housing in Singapore, and it could also be utilised in other similar urban areas for future electricity consumption forecasting.
... The study concluded that the MLP approach was the most accurate, while SVM came in second. Dong et al. (2005) predicted monthly electricity usage in tropical locations. They showed that SVM was a good predictor of electricity use based on three years of data analysis predicting cooling loads for office buildings. ...
Improving load forecasting is becoming increasingly crucial for power system management and operational research. Disruptive influences can seriously impact both the supply and demand sides of power. This work examines the impact of the coronavirus on power usage in two US states from January 2020 to December 2020. A wide range of machine learning (ML) algorithms and ensemble learning are employed to conduct the analysis. The findings showed a surprising increase in monthly power use changes in Florida and Texas during the COVID-19 pandemic, in contrast to New York, where usage decreased over the same period. In Texas, the quantity of power usage rises from 2% to 6% practically every month, except for September, when it decreased by around 1%. For Florida, except for May, which showed a fall of roughly 2.5%, the growth varied from 2.5% to 7.5%. This indicates the need for more extensive research into such systems and the applicability of adopting groups of algorithms in learning the trends of electric power demand during uncertain events. Such learning will be helpful in forecasting future power demand changes due to especially public health-related scenarios.
... Many of the present-day techniques lack efficient feature selection methods and suffer loss of accuracy Zhiyong Du, et al. proposed a method [11]. Dong, et al. [12] proposed a Support Vector machine technique for identifying the useful features of time series data and prepared a model for predicting energy demand of buildings in tropical regions. Another model was built on the same lines for predicting the annual energy demand of a building using heat transfer coefficients and SVM Xiao, Z. et al. [13]. ...
Climate change, as known, is the dangerous environmental effect we are going to face in the near future and electricity contributes the majority of its part in overcoming climate change as per the trends. Usage of electricity is widely increasing all over the world mainly as an alternative to the use of fossil fuels. In households the usage is rapidly increasing day by day, owing to the increase in the number of devices running on electricity. As we have observed mainly after the relaxation of the lockdown the bills received by households, especially in cities were unhappy and have left most of the people aghast. It is evident that users have no idea about the power they consume. In this work, a model to forecast the electricity bill of household users based on the previous trends and usage patterns by making use of machine learning techniques has been proposed. The historical data of the user is studied and the learning is done iteratively to improve the accuracy of the model. The model can then be used to forecast the consumption beforehand.
Residential building energy consumption is a significant contributor to greenhouse gas emissions. Accurate prediction of total energy use in residential buildings holds vital importance in the context of energy management. In this paper, we propose a data-driven approach using machine learning (ML) models to predict the total energy consumption of a low-energy house based on indoor and outdoor environmental conditions using data from a house located in Belgium. Four ML(ML) models, including Extreme Gradient Boosting (XGBoost), Random Forest (RF), Decision Tree (DT), and Support Vector Machine (SVM), were trained and tested to evaluate their performance in predicting energy consumption. The results of our study demonstrate that the XGBoost model outperforms all other models used, with a coefficient of determination R2 of 61%, a Root Mean Square Error (RMSE) of 65.28, a Mean Absolute Error (MAE) of 29.81, and a Mean Absolute Percentage Error (MAPE) of 28.55 on the testing set. The findings from this study demonstrate the accurate forecasting of energy consumption by accounting for the non-linear dependencies between environmental conditions and total energy consumption, which can aid in making informed decisions towards the reduction of power usage, enhancement of energy efficiency, and achieving cost savings.
Accurate and efficient prediction of electric water boiler (EWB) energy consumption is significant for energy management, effective demand response, cost minimisation, and robust control strategies. Adequate tracking and prediction of user behaviour can enhance renewable energy minigrid (REMD) management. Fulfilling these demands for predicting the energy consumption of electric water boilers (EWB) would facilitate the establishment of a new framework that can enhance precise predictions of energy consumption trends for energy efficiency and demand management, which necessitates this state-of-the-art review. This article first reviews the factors influencing the prediction of energy consumption of electric water boilers (EWB); subsequently, it conducts a critical review of the current approaches and methods for predicting electric water boiler (EWB) energy consumption for residential building applications; after that, the performance evaluation methods are discussed. Finally, research gaps are ascertained, and recommendations for future work are summarised.
Artificial neural network (ANN) has acquired noticeable interest from the research community to handle complex problems in Construction and Building engineering (CB). This interest has led to an enormous amount of scientific publications in diverse CB domains over the last three decades. This study introduces a scientometric review to quantitatively explore and visually map the development pathways and trends of ANN-CB literature. Via the Web of Science (WoS) database, 2406 peer-reviewed journal articles are identified and included for analysis as follows. First, the publication growth over time is inspected and evaluated. Second, the collaboration patterns between key contributors (researchers, countries, and organizations) are explored and mapped using the co-authorship analysis. Third, the key sources’ productivity and influence are explored and mapped using the direct citation analysis. Fourth, the publications clusters and research themes are analyzed and visualized via the keyword co-occurrence analysis and document trend topics mapping. The study outcomes include but are not limited to i) recognizing pioneer ANN-CB researchers for future collaboration opportunities, ii) identifying reliable sources of information or suitable ones for publishing new ANN-CB works, and iii) fostering probable academic partnerships with the leading ANN-CB organizations. These outcomes help researchers to comprehend ANN-CB literature and direct research policy-makers and editorial boards to adopt the promising ANN-CB themes for further research and development.
As the world grapples with the challenges posed by climate change and depleting energy resources, achieving sustainability in the construction and operation of buildings has become a paramount concern. The construction and operation of buildings account for a substantial portion of global energy consumption and carbon emissions. Hence, the accurate prediction of building energy consumption is indispensable for reducing energy waste, minimizing greenhouse gas emissions, and fostering sustainable urban development. The aspiration to achieve predicted outcomes with remarkable accuracy has emerged as a pivotal objective, coinciding with the burgeoning popularity of deep learning techniques. This paper presents an auto-evaluation model for building energy consumption prediction via Long Short-Term Memory with modified Kalman filtering (LSTM-MKF). Results gleaned from data validation activities evince a notable transformation—a reduction of the maximal prediction error from an initial 83% to a markedly ameliorated 24% through the intervention of the proposed model. The LSTM-MKF model, a pioneering contribution within this paper, clearly exhibits a distinct advantage over the other models in terms of predictive accuracy, as underscored by its superior performance in all three key metrics, including mean absolute error, root mean square error, and mean square error. The model presents excellent potential as a valuable tool for enhancing the precision of predictions of building energy consumption, a pivotal aspect in energy efficiency, smart city development, and the formulation of informed energy policy.
Multiple tools exist for separately simulating and estimating the parameters of system dynamics models. Artificial intelligence (AI) has been increasingly used to estimate the parameters of system dynamics models. The development of modeling tools and advanced environments has resulted in great benefits to the community at large. The incorporation of AI tools into system dynamics presents opportunities for expanding on current decision‐making methods. As systems become complex, the need to incorporate evidence‐based data‐driven methods increases. By integrating system dynamics tools and facilitating AI and system dynamics simulation in an integrated environment, model parameters can be estimated with the latest data, and the integrity of the model can be retained effectively. This provides an advantage to the efficiency and capabilities of the system dynamics model and its analysis. This paper presents a general methodology to incorporate regression AI into system dynamics models for simulation and analysis. To demonstrate the validity of the methodology, a case study involving a susceptible‐infected‐recovered model and empirical data from the COVID‐19 pandemic is performed using support vector machines (SVMs), artificial neural networks (ANNs), and random forests.
The endocrine system is responsible for secreting and controlling hormones crucial in regulating key body activities. However, endocrine disruptors or endocrine-disrupting chemicals (EDCs) can harm human health and well-being by interfering with this complex process. This report seeks to assess the present state of understanding about endocrine disruptors in China, including their origins, impacts, and obstacles, and to provide actionable recommendations for reducing exposure and mitigating negative effects. Strong negative correlations between ANOE and rural ecological compensation (REC) and a negative correlation between ANOE and forest coverage (FC) were found in this analysis of the relationships between agricultural nitrous oxide emissions (ANOE), agricultural methane emissions (AME), and land use and land cover variables (LUPC). Just as LUPC is significantly inversely related to FC, AME is positively related. The team uses a gradient-boosted model (GBM) with a Gaussian loss function and fine-tunes the model's parameters to achieve optimal performance and reliable prediction results. With a relative relevance score of 90.36 for ANOE and 67.64 for AME, the analysis shows that LUPC is the most important factor in influencing emission levels. This study aims to increase knowledge of endocrine disruptors' potential advantages and disadvantages in outdoor exercise. The study aims to aid in preventing and managing many diseases and disorders caused by hormonal imbalances or disruptions by examining the origins, effects, and potential mitigation of these substances during outdoor activity. Safe and healthful outdoor exercise is promoted by the study's efforts to discover and implement effective and sustainable solutions to decrease emissions and exposure to endocrine disruptors. This comprehensive study aims to promote a healthier and more sustainable environment for individuals engaging in outdoor exercise by synthesizing current knowledge, providing practical recommendations, and emphasizing the importance of awareness and action.
Indoor temperature prediction is an essential component of building control and energy-saving. Although existing indoor temperature prediction frameworks have achieved remarkable progress, they struggle to achieve high performance due to information, method, application, and sim-to-real gaps. Aiming to fill these gaps, we propose a novel deep-learning framework for short-term indoor temperature prediction in multi-zone buildings. In particular, we expand the sensing information and formulate the multi-zone indoor temperature prediction (MITP) problem. To improve the prediction performance, we employ information fusion and encoder–decoder architecture to the MITP problem and propose MITP-Net. We set up 11 ablation experiments to compare the prediction performance of relative frameworks. To evaluate frameworks’ performance, we publicly release a dataset including 2-week real operating data in a multi-zone office with a 1-min sampling interval (829,440 digits in total). Compared with existing deep-learning frameworks, MITP-Net significantly raises the prediction accuracy and can flexibly adjust the lengths of input and prediction sequences for different requirements. We provide the usage steps of MITP-Net and publish the operating data and codes on the GitHub repository: https://github.com/XingTian1994/MITP-Net.
Regression analysis is used to define algebraic expressions which can be used to study the effects of various configuration parameters on building energy performance. The DOE-2.1B energy analysis simulation program was used in conjunction with a prototypical building modules to generate a large data base for five geographic locations. Configuration variables parameterized included orientation, wall and roof conductance, glazing properties of windows and skylights (area, U-value, shading coefficient, visible transmittance, well depth) and installed lighting power. Incremental effects due to external shading (overhangs) and daylighting were also examined. The modular concept used in the building definition permits generalization to configurations other than that analyzed and therefore the results can be used in studying building design trade-offs.
A novel method for regression has been recently proposed by Vapnik
et al. (1995, 1996). The technique, called support vector machine (SVM),
is very well founded from the mathematical point of view and seems to
provide a new insight in function approximation. We implemented the SVM
and tested it on a database of chaotic time series previously used to
compare the performances of different approximation techniques,
including polynomial and rational approximation, local polynomial
techniques, radial basis functions, and neural networks. The SVM
performs better than the other approaches. We also study, for a
particular time series, the variability in performance with respect to
the few free parameters of SVM
A novel type of learning machine called support vector machine (SVM) has been receiving increasing interest in areas ranging from its original application in pattern recognition to other applications such as regression estimation due to its remarkable generalization performance. This paper deals with the application of SVM in financial time series forecasting. The feasibility of applying SVM in financial forecasting is first examined by comparing it with the multilayer back-propagation (BP) neural network and the regularized radial basis function (RBF) neural network. The variability in performance of SVM with respect to the free parameters is investigated experimentally. Adaptive parameters are then proposed by incorporating the nonstationarity of financial time series into SVM. Five real futures contracts collated from the Chicago Mercantile Market are used as the data sets. The simulation shows that among the three methods, SVM outperforms the BP neural network in financial forecasting, and there are comparable generalization performance between SVM and the regularized RBF neural network. Furthermore, the free parameters of SVM have a great effect on the generalization performance. SVM with adaptive parameters can both achieve higher generalization performance and use fewer support vectors than the standard SVM in financial forecasting.
The sigmoid kernel was quite popular for support vector machines due to its origin from neural networks. However, as the kernel matrix may not be positive semidefinite (PSD), it is not widely used and the behavior is unknown. In this paper, we analyze such non-PSD kernels through the point of view of separability. Based on the investigation of parameters in different ranges, we show that for some parameters, the kernel matrix is conditionally positive definite (CPD), a property which explains its practical viability. Experiments are given to illustrate our analysis. Finally, we discuss how to solve the non-convex dual problems by SMO-type decomposition methods. Suitable modifications for any symmetric non-PSD kernel matrices are proposed with convergence proofs.
Though area in office buildings has expanded significantly in Hong Kong in the decade to 2013, increase in cooling energy use has been rather moderate. The factors and policy instruments that contributed to the impressive enhancement in energy usage efficiency are not available in published literature. Data of sixty-six office developments were examined to evaluate the role of policy instruments in bringing about the visible improvement. Numerical analyses reveal that use of chiller units with higher efficiency and fresh water cooling tower (FWCT) systems have contributed significantly to reduce cooling energy use of office space in Hong Kong. It is important to note that policies promulgated have played a key role in encouraging designers of office space to opt for energy efficient systems. The BEAM Plus certification for all public buildings has contributed the most (efficacy = 0.09) to the success in moderation of electricity consumption for cooling of office space. The FWCT scheme and the Building Energy Codes have had efficacies amounting to 0.076 and 0.026, respectively. Interestingly, the new OTTV requirement, used for regulation and monitoring of thermal performance of building envelopes, has had a negative rate of efficacy (0.018), implying that the government needs to review its appropriateness.
A baselining methodology is crucial to verify savings from energy conservation programs, to determine progress toward preset energy-efficiency goals, and to implement performance-based shared energy savings contracts. This paper discusses the various issues involved and structures a formal baselining methodology at the facility level or whole-building level when monthly utility bills are available. Normalizing annual energy use for changes in the conditioned area and in the number of occupants as well as correcting for increases in the connected load (a phenomenon referred to as 'creep') are also discussed. The various functional forms assumed by the regression models with outdoor dry-bulb temperature as the only regressor variable are described in the framework of two widely used energy modeling software packages. An important aspect of this paper is the development of prediction uncertainty bands for the regression model, both at the monthly level and at the annual level, an issue of key importance if one wishes to reach statistically meaningful conclusions regarding month-to-month and year-to-year changes in observed energy use with respect to the baseline year. Finally, a statistical method is described by which one can overcome model identification limitations in case the read dates of the utility bills are not known. A companion paper presents the results of applying this baselining methodology to eight Army installations nationwide.
This paper compares methods to verify post-retrofit energy savings using available data on existing buildings. Common methods, including the 65°F-base degree-day method, the bin method, linear change-point regression, and neural networks, were assessed. A new method for predicting building energy use, based on genetic programming, was developed. All methods were experimentally evaluated using data collected from two existing buildings. These two buildings did not have any retrofits installed, so the known energy savings was zero. The evaluation procedures are described and the results compared.
In the history of research of the learning problem one can extract four periods that can be characterized by four bright events: (i) Constructing the first learning machines, (ii) constructing the fundamentals of the theory, (iii) constructing neural networks, (iv) constructing the alternatives to neural networks.
Accurate modeling of hourly heating and cooling energy use in commercial buildings can be achieved by a Generalized Fourier Series (GFS) approach involving weather variables such as dry-bulb temperature, specific humidity and horizontal solar flux. However, there are situations when only temperature data is available. The objective of this paper is to (i) describe development of a variant of the GFS approach which allows modeling both heating and cooling hourly energy use in commercial buildings with outdoor temperature as the only weather variable and (ii) illustrate its application with monitored hourly data from several buildings in Texas. It is found that the new Temperature based Fourier Series (TFS) approach (1) provides better approximation to heating energy use than the existing GFS approach, (ii) can indirectly account for humidity and solar effects in the cooling energy use, (iii) offers physical insight into the operating pattern of a building HVAC system and (iv) can be used for diagnostic purposes.
This paper overviews some applications of neural networks (NNs) to estimate energy and demand savings from retrofits of commercial buildings. First, a brief background information on NNs is provided. Then, three specific case studies are described to illustrate how and when NNs can be used successfully to determine energy savings due to the implementation of various energy conservation measures in existing commercial buildings.
An empirical or regression modeling approach is simple to develop and easy to use compared to detailed hourly simulations of energy use in commercial buildings. Therefore, regression models developed from measured energy data are becoming an increasingly popular method for determining retrofit savings or identifying operational and maintenance (O and M) problems. Because energy consumption in large commercial buildings is a complex function of climatic conditions, building characteristics, building usage, system characteristics and type of heating, ventilation, and air conditioning (HVAC) equipment used, a multiple linear regression (MLR) model provides better accuracy than a single-variable model for modeling energy consumption. Also, when hourly monitored data are available, an issue which arises is what time resolution to adopt for regression models to be most accurate. This paper addresses both these topics. This paper reviews the literature on MLR models of building energy use, describes the methodology to develop MLR models, and highlights the usefulness of MLR models as baseline models and in detecting deviations in energy consumption resulting from major operational changes. The paper first develops the functional basis of cooling energy use of two commonly used HVAC systems: dual-duct constant volume (DDCV) nd dual-duct variable air volume (DDVAV). Using these functional forms, the cooling energy consumption in five large commercial buildings located in central Texas were modeled at monthly, daily, hourly, and hour-of-day (HOD) time scales. Compared to the single-variable model (two-parameter model with outdoor dry-bulk as the only variable). MLR models showed a decrease in coefficient of variation (CV) between 10% to 60%, with an average decrease of about 33%, thus clearly indicating the superiority of MLR models. Although the models at the monthly time scale had higher coefficient of determination (R²) and lower CV than daily, hourly, and HOD models, the daily and HOD models proved more accurate at predicting cooling energy use.
The methodology for baseline building energy consumption is well established for energy saving calculation in the temperate zone both for performance-based energy retrofitting contracts and measurement and verification (M&V) projects. In most cases, statistical regression models based on utility bills and outdoor dry-bulb temperature have been applied to baseline monthly and annual whole building energy use. This paper presents a holistic utility bills analysis method for baseline whole building energy consumption in the tropical region. Six commercial buildings in Singapore were selected for case studies. Correlationships between the climate data, which are monthly mean outdoor dry-bulb temperature (T0), relative humidity (RH) and global solar radiation (GSR), and whole building energy consumption are derived. A deep prediction study based monthly mean outdoor dry-bulb temperature (T0) and whole building energy consumption is stated. The result shows that variations of the energy consumption in most of these buildings are contributed by T0 and can be well predicted at 90% confidence level only with it. The analysis of such kind of model is especially useful for building managers, owners and ESCOs to track and baseline energy use during pre-retrofit and post-retrofit periods in the tropical condition.
Regression analysis defines algebraic expressions that can be used to study the effects of various configurational parameters on building energy performance. The DOE-2.1B energy-analysis simulation program was used in conjunction with prototypical building modules to generate a large data base for five geographic locations. Configurational variables parameterized included orientation, wall and roof conductance, glazing properties of windows and skylights (area, U-value, shading coefficient, visible transmittance, well depth), and installed lighting power. Incremental effects due to external shading (overhangs) and daylighting were also examined. A constant volume variable temperature secondary HVAC system was attached to each zone in the model, and the individual zone coil loads were used as a basis for showing the versatility of the regression techniques. The modular concept permits generalization of the results to buildings of arbitrary floor/glazing areas, and the resulting regression expression conveniently categorizes those components of importance in building energy use.
This paper describes a study conducted in the lighting sector of office buildings as a part of a broader research study aimed at developing building codes for Sri Lanka addressing lighting as well as thermal comfort in order to optimise the use of electricity within these buildings. The study covered different tasks performed in office buildings and the optimum lighting levels required to perform these tasks in the office environment in Sri Lanka. Also, it included assessing the visual performance of people involved in different activities under varying illumination levels in a controlled environment and a comparison of these optimum lighting levels with international standards. It can be seen that the required optimum lighting levels are generally lower in Sri Lanka in comparison to specified standard levels, and this scenario is likely to be similar in other developing countries too. These findings clearly emphasise the need to adopt lighting standards most appropriate to local conditions, in turn helping improve the energy efficiency within buildings.
LIBSVM is a library for support vector machines (SVM). Its goal is to help users to easily use SVM as a tool. In this document, we present all its imple-mentation details. For the use of LIBSVM, the README file included in the package and the LIBSVM FAQ provide the information.
Support Vector Machines are used for time series prediction and compared to radial basis function networks. We make use of two different cost functions for Support Vectors: training with (i) an e insensitive loss and (ii) Huber's robust loss function and discuss how to choose the regularization parameters in these models. Two applications are considered: data from (a) a noisy (normal and uniform noise) Mackey Glass equation and (b) the Santa Fe competition (set D). In both cases Support Vector Machines show an excellent performance. In case (b) the Support Vector approach improves the best known result on the benchmark by a factor of 29%.
We report a novel possibility for extracting asmall subset of a data base which contains allthe information necessary to solve a given classificationtask: using the Support Vector Algorithmto train three different types of handwrittendigit classifiers, we observed that these typesof classifiers construct their decision surface fromstrongly overlapping small (? 4%) subsets of thedata base. This finding opens up the possibilityof compressing data bases significantly by disposingof the...
Support vector machine (SVM) is a popular technique for classi�cation.
However, beginners who are not familiar with SVM often get unsatisfactory
results since they miss some easy but signi�cant steps. In this guide, we propose
a simple procedure, which usually gives reasonable results.
Measured energy savings promote and sustain energy conservation retrofits by verifying the success of retrofits, determining pay-back schedules, guiding the selection of future retrofits and identifying opportunities for further savings. This dissertation develops a methodology to measure retrofit energy savings and the uncertainty of the savings in commercial buildings. The functional forms of empirical models of cooling and heating energy use in commercial buildings are derived from an engineering analysis of constant-air-volume and variable-air-volume HVAC systems. One, two, three and four parameter, temperature-dependent regression models are proposed to model baseline energy use. Retrofit savings are measured as the difference between the baseline energy use project by the models and the measured post-retrofit energy use. A hybrid ordinary least squares/autoregressive method is developed to determine the uncertainty of the predicated energy use and savings. The annual predictive ability of models based on pre-retrofit data sets of less than a full year is investigated. The energy delivery efficiency is introduced to measure the efficiency of air-side systems at meeting the net building load. A preliminary investigation of the use of artificial neural network models to measure savings is presented. The methodology is demonstrated on case study examples using software specifically developed for the analysis of commercial building energy use.
Support vector machines (SVMs) with the gaussian (RBF) kernel have been popular for practical use. Model selection in this class of SVMs involves two hyperparameters: the penalty parameter C and the kernel width sigma. This letter analyzes the behavior of the SVM classifier when these hyperparameters take very small or very large values. Our results help in understanding the hyperparameter space that leads to an efficient heuristic method of searching for hyperparameter values with small generalization errors. The analysis also indicates that if complete model selection using the gaussian kernel has been conducted, there is no need to consider linear SVM.
We investigate practical selection of hyper-parameters for support vector machines (SVM) regression (that is, epsilon-insensitive zone and regularization parameter C). The proposed methodology advocates analytic parameter selection directly from the training data, rather than re-sampling approaches commonly used in SVM applications. In particular, we describe a new analytical prescription for setting the value of insensitive zone epsilon, as a function of training sample size. Good generalization performance of the proposed parameter selection is demonstrated empirically using several low- and high-dimensional regression problems. Further, we point out the importance of Vapnik's epsilon-insensitive loss for regression problems with finite samples. To this end, we compare generalization performance of SVM regression (using proposed selection of epsilon-values) with regression using 'least-modulus' loss (epsilon=0) and standard squared loss. These comparisons indicate superior generalization performance of SVM regression under sparse sample settings, for various types of additive noise.
This paper is an extended version of [12]. Generic author design sample pages 2000/07/31 03:05
The Support Vector (SV) method was recently proposed for estimating regressions, constructing multidimensional splines, and solving linear operator equations [Vapnik, 1995]. In this presentation we report results of applying the SV method to these problems. 1 Introduction The Support Vector method is a universal tool for solving multidimensional function estimation problems. Initially it was designed to solve pattern recognition problems, where in order to find a decision rule with good generalization ability one selects some (small) subset of the training data, called the Support Vectors (SVs). Optimal separation of the SVs is equivalent to optimal separation the entire data. This led to a new method of representing decision functions where the decision functions are a linear expansion on a basis whose elements are nonlinear functions parameterized by the SVs (we need one SV for each element of the basis). This type of function representation is especially useful for high dimensional...
We report a novel possibility for extracting a small subset of a data base which contains all the information necessary to solve a given classification task: using the Support Vector Algorithm to train three different types of handwritten digit classifiers, we observed that these types of classifiers construct their decision surface from strongly overlapping small ( 4%) subsets of the data base. This finding opens up the possibility of compressing data bases significantly by disposing of the data which is not important for the solution of a given task. In addition, we show that the theory allows us to predict the classifier that will have the best generalization ability, based solely on performance on the training set and characteristics of the learning machines. This finding is important for cases where the amount of available data is limited. Introduction in: U. M. Fayyad and R. Uthurusamy (eds.): Proceedings, First International Conference on Knowledge Discovery ...
Special issue devoted to measuring energy savings, the Princeton scorekeeping method (PRISM)
- Fels
M. Fels, Special issue devoted to measuring energy savings, the
Princeton scorekeeping method (PRISM), Energy and Buildings 9
(1-2) (1986).
Regime signalling techniques for nonstationary time series forecasting
- Doss
R. Doss, Z. Obradovic, Regime signalling techniques for nonstationary time series forecasting, Neurovest Journal 4 (5) (1998) 7-15.
Neural Network FAQ. Periodic posting to the Usenet news group comp.ai.neural-nets available
- W S Sarle
W.S. Sarle, 1997. Neural Network FAQ. Periodic posting to the Usenet news group comp.ai.neural-nets available: ftp://ftp.sas.com/pub/ neural/FAQ.html.
Applying neural networks to model monthly energy consumption of commercial buildings in Singapore
- B Dong
- S E Lee
- M H Sapar
- H S Sun
B. Dong, S.E. Lee, M.H. Sapar, H.S. Sun, Applying neural networks to
model monthly energy consumption of commercial buildings in
Singapore, in: Proceedings of International Conference on Control,
Automation and System, Thailand, 2004.
Energy efficiency of office buildings in Singapore
- Lee
Applying neural networks to model monthly energy consumption of commercial buildings in Singapore
- Dong