Fig 3 - uploaded by Kamran Javed
Content may be subject to copyright.
Source publication
Prognostics is a core process of Prognostics & Health Management (PHM) discipline, that estimates the Remaining Useful Life (RUL) of a degrading machinery to optimize its service delivery potential. However, machinery operates in a dynamic environment and the acquired condition monitoring data are usually noisy and subject to high level of uncertai...
Context in source publication
Context 1
... M F E is the mean forecast error between the actual values of T S and the predicted ones (see details in [18]). Predictability has an exponential form (Fig. 3) and is inversely proportional to M F E. A T S can be considered predictable if predictability coefficient is between 0.5 and 1. The aim of this measure is to reconsider the learning step of data- driven approach by considering both "feature selection" and "prognostics modeling" as interrelated. ...
Similar publications
![Table 1 : Detailed description of NASA bearing dataset [8]](profile/Pradeep-Gopalakrishnan/publication/328445812/figure/tbl1/AS:684735462391810@1540265064513/Detailed-description-of-NASA-bearing-dataset-8_Q320.jpg)
Condition monitoring is one of the routine tasks in all major process industries. The mechanical parts such as a motor, gear, bearings are the major components of a process industry and any fault in them may cause a total shutdown of the whole process, which may result in serious losses. Therefore, it is very crucial to predict any approaching defe...
Citations
... FCM is trained in an unsupervised way and aims to find similarities within the data. In the literature, FCM has been successfully implemented for the purpose of health state division Javed, Gouriveau, and Zerhouni 2015;Wang et al. 2019). The main advantage of FCM algorithm over hard clustering algorithms is its ability to allow gradual memberships of data points to clusters measured as degrees in the interval [0, 1]. ...
In many settings, fleets of assets must perform series of missions with in-between finite breaks. For such fleets, a widely used maintenance strategy is the fleet selective maintenance (FSM). Under resource constraints, the FSM problem selects an optimal subset of feasible maintenance actions to be performed on a subset of components to minimise the maintenance cost while ensuring high system reliability during the upcoming mission. The majority of extant FSMP models are focussed on traditional physics-based reliability models. With recent advances in Machine Learning (ML) and Deep Learning (DL) algorithms, data-driven methods have shown accuracy in predicting remaining useful life (RUL). This paper proposes a predictive FSM strategy for fleets of complex and large multi-component systems. It relies on a concurrent ML/DL and optimisation approach where a clustering algorithm is used to determine the health states of components and a probabilistic RUL prognostics model is used for component reliability assessment. An optimisation model is developed to solve the predictive FSM problem to ensure high reliability of all systems within the fleet. An efficient two-phase solution approach is developed to solve this complex optimisation problem. Numerical experiments show the validity of the approach and highlight the improved maintenance plans achieved.
... Developing a precise model that can accurately predict the dynamic behavior of systems or objects, especially complex nonlinear behaviors [23][24][25]. Specifically, Javed et al [26] integrated extreme learning machine and fuzzy clustering to describe the degradation evolution of a turbofan engine by estimating and predicting states. Salmela et al [27] used recursive neural networks to predict ultrafast nonlinear dynamics in optical fibers, solving the problem of complex optimization of physical models. ...
Acoustic weighing is a promising method for non-contact mass measurement of tiny objects as it avoids contamination and contact losses. However, due to the highly nonlinear nature of the acoustic field, some parameters of the mechanism model of acoustic weighing cannot be accurately simulated, thereby reducing the accuracy of acoustic weighing. To improve the accuracy of acoustic weighing, we propose an acoustic weighing method based on oscillating signals and feature enhancement network. Firstly, to drive the object oscillation and collect oscillation data, an acoustic levitation-based data acquisition system is constructed. Then, to break the limitations of the mechanism model, a feature enhancement network named CNN-BiLSTM-SE is proposed, which directly establishes the correlation between oscillating signals and actual mass. Finally, these data are used to train and test the proposed network model, validating the effectiveness of the model. Experimental results show that the method achieves high accuracy in measuring object mass, following the actual measurements with remarkable consistency. In addition, our approach is also suitable for acoustic weighing of small and sensitive objects, opening up new perspective for the study and application of nonlinear acoustic systems.
... (3) In multivariate time series prediction problems, considering the variability of different samples under each feature and the similarity between samples helps to improve the prediction performance and accuracy. Compared with BPNN [49], Elman [50], Javed et al.'s model [51], Zheng et al.'s model [52], MBBO-ESN [28] and MBSSA-ESN [29], the MBSSA-Bi-AESN model developed in this paper achieves the best classification accuracy in terms of results because it considers both the variability of features of different samples and the similarity between samples. ...
... (2) In the multivariate time series prediction process, the similarity between objects needs to be considered. However, BPNN [49], Elman [50], Javed et al.'s model [51] and Zheng et al.'s model [52], MBBO-ESN [28] and MBSSA-ESN [29] Fig. 12 ROC curve for "Spectfheart" and "Eastwest" all consider all the objects together, resulting in redundant features appearing or the selected subset of features is not optimal. In contrast, the proposed MBSSA-Bi-AESN model in this essay applies the clustering idea to the prediction process, which can eliminate redundant features and also make the selected subset of features optimal. ...
... (3) When performing feature selection operations, it is important to fully consider the variability of different objects under each feature for the prediction effectiveness of subsequent predictions. In these prediction models, compared to BPNN [49], Elman [50], Javed et al.'s model [51] and Zheng et al.'s model [52], MBBO-ESN [28] and MBSSA-ESN [29], the MBSSA-Bi-AESN model fully considered the actual differences of different objects under the corresponding features, and combined with the clustering results to give the features corresponding weights. This makes the selected subset of features more accurate, thus improving the prediction accuracy and classification effect of the established MBSSA-Bi-AESN model. ...
In the era of big data, the demand for multivariate time series prediction has surged, drawing increased attention to feature selection and neural networks in machine learning. However, certain feature selection methods neglect the alignment between actual data sample differences and clustering results, while neural networks lack automatic parameter adjustment in response to changing target features. This paper presents the MBSSA-Bi-AESN model, a Bi-directional Adaptive Echo State Network that utilizes the modified salp swarm algorithm (MBSSA) and feature selection to address the limitations of manually set parameters. Initial feature subset selection involves assigning weights based on the consistency of clustering results with differences. Subsequently, the four critical parameters in the Bi-AESN model are optimized using MBSSA. The optimized Bi-AESN model and selected feature subset are then integrated for simultaneous model learning and optimal feature subset selection. Experimental analysis on eight datasets demonstrates the superior prediction accuracy of the MBSSA-Bi-AESN model compared to benchmark models, underscoring its feasibility, validity, and universality.
... Data-driven RUL prediction methods aim to construct a degradation model based on the nonlinear relationship between sensor data and equipment RUL labels. These methods can be further divided into traditional machine-learning-based techniques, such as autoregressive models (ARs) [7], support vector machines (SVMs) [8], extreme learning machines (ELMs) [9], and random forests (RFs) [10], and deep learning-based methods. Although traditional machine learning techniques have been widely applied in RUL prediction, their accuracy is often limited by manual expertise and feature engineering. ...
Artificial intelligence (AI) methods have significantly improved the accuracy of RUL estimation. However, these methods often neglect issues such as inadequate deep feature mining and loss of critical information during prediction. This study proposes a baseline similarity attention-based dual-channel feature fusion network (BSA-DCFF) for RUL prediction. Firstly, the input data undergoes preliminary feature extraction using two independent channels to obtain hidden states at each time step. The feature extraction process for each channel operates independently, ensuring thorough data extraction. Secondly, the baseline similarity attention mechanism (BSA) is applied independently to both channels. By employing feature fusion, it compensates for missing information during the extraction process and enhances degraded features. Finally, the features extracted from the two channels are concatenated and fed into a fully connected subnetwork for RUL prediction. To further improve the accuracy of the potential representation obtained by the BSA structure, additional model optimization is performed by reconstructing the joint loss function. The proposed method is evaluated through an experiment on a publicly available dataset to evaluate its effectiveness.
... NASA Ames Research Center has released the engine operation-failure simulation dataset to facilitate researchers in exploring and developing data-driven RUL prediction technology. The data set is generated by conducting a large number of engine performance and degradation simulation experiments on NASA's civil aviation propulsion system and simulation platform CMAPSS (Commercial Modular Aero-Propulsion System Simulation) according to the engine damage expansion modeling proposed in (Saxena et al., 2008), which provides excellent convenience for researchers to test and validate data-driven RUL prediction methods, so it is widely used in many studies (Chao et al., 2021;García Nieto et al., 2015;Heimes, 2008;Javed et al., 2015). ...
Changes in the performance of an aircraft system will straightforwardly affect the safe operation of the aircraft, and the technical requirements of Prognostics and Health Management (PHM) are highly relevant. Remaining Useful Life (RUL) prediction, part of the core technologies of PHM, is a cutting-edge innovation being worked on lately and an effective means to advance the change of upkeep support mode and work on the framework's security, unwavering quality, and economic reasonableness. This paper summarizes a detailed preliminary literature review and comparison of different prognostic approaches and the forecasting methods' taxonomy, the methodology's details, and provides its application to aircraft systems. It also provides a brief introduction to the predictive maintenance concept and condition-based maintenance (CBM). This article uses several predictive models to predict RUL and classifies conventional regression algorithms according to the similarity in function and form of the algorithms. More classical algorithms in each category are selected to compare the prediction results, and finally, the combined effects of the RUL prediction are obtained by weighted fusion, accuracy, and compatibility. The performance of the proposed models is assessed based on evaluations of RUL acquired from the hybrid and individual predictive models. This correlation depends on the most current prognostic metrics. The outcomes show that the proposed strategy develops precision, robustness, and adaptability. Hence, the work in this paper shall enrich the advancement of predictive maintenance and modern innovation of prognostic development.
... Since all the evidence information is included in the first scenario and different degradation statuses are combined in this scenario, the prognosis results are more accurate and have fewer errors. For instance, the score criterion in the first scenario is equal to 398.1, which is smaller than (better) the second scenario's score of 832.6 (Javed, Gouriveau, and Zerhouni 2015;Fu et al. 2021). As for the performance criterion, it is 73.1 under the first scenario, which is larger than (better) the second scenario's performance of 53.7 (Bektas et al. 2019;Deng et al. 2020). ...
... For the proposed model, operating condition one has been used to compare with the existing method. Javed, Gouriveau, and Zerhouni (2015) used an artificial neural network method without taking into account the similarity between the training and test datasets. The results of the final prognosis are obtained based on the average outputs of the neural network for each training dataset. ...
... Extreme Learning Machine and Fuzzy Clustering (Javed, Gouriveau, and Zerhouni 2015) 1046 48 Case-based Reasoning (Ramasso 2014) N/A 53 Instance-Based Learning (Khelif et al. 2014) N/A 54 Support Vector Regression (Khelif et al. 2017) 448 70 times series memory auto-encoder (Fu et al. 2021) 200 59 CatBoost (Deng et al. 2020) 398.7 N/A Hamiltonian Monte Carlo (Benker et al. 2021) 427 N/A Rao-Blackwellized Particle Filter (Cai et al. 2020) 679.19 N/A Bidirectional recurrent neural network (Yu, Kim, and weight for the mean RUL in each cluster. Finally, the RUL is obtained through the fusion of information. ...
In prognostics and health management, the system’s degradation condition assessment and
corresponding remaining useful life prediction are the most important tasks. Both of these
processes are heavily dependent on information gathered by multiple sensors, which eventually
causes data fusion-related complex problems. Typically, sensor information contains
the speed, pressure, temperature, and similar other types of various system data. These systems’
data obtained through sensors can be utilized as a part of the evidence in the evidence-
based estimation method. In this work, an artificial intelligence-based novel
framework for estimating the remaining useful life using data fusion has been presented.
The Dempster–Shafer extended theory is adopted for sensor information modeling and data
fusion. Besides, two different scenarios are introduced to determine the similarity between
the studied system and the available evidence. As a case study, the turbofan dataset is
demonstrated to assess the proposed method. Based on the results, our integrated proposed
method performs very competitively compared with the existing methods based on
standard scores and performance criteria.
... Given the diversity of prognostic models, the definition of metrics for assessing and comparing the performance of prognostic algorithms can be of defining importance in decision making [14,[43][44][45][46][47][48]. Recent papers review such metrics [10,49]. ...
... Machine learning (ML) based models seek to learn the degradation patterns of machines with ML techniques from existing observations, rather than build physics models or statistical models. Machine learning approach such as neural networks [12][13][14][15][16][17][18][19], have strong potential to be used for diagnostics and prognostics areas. They are capable of solving the diagnostics and prognostics problem of complex mechanical systems whose degradation processes are difficult to interconnect with physics or statistical models. ...
... Using Bayes' theorem and applying the results of Eqs. (18) and (19) we obtain the proper weights: ...
The number of timely diagnoses based on condition monitoring data is increasing with the growing usage of monitoring systems. In most of the methods used in these systems, a pre-established fault detection threshold is needed, while there are no specific limit values or thresholds in many cases, especially when the machine is unique. Also, in most actual applications, due to the kind of process and harsh environment, the noise inherent in the observed process exhibits non-Gaussian characteristics, making it a challenging task for diagnostics based on condition monitoring (CM) data. Therefore, this paper introduced a robust methodology based on the switching maximum correntropy Kalman filter (SMCKF) to address the mentioned problems (threshold and online diagnostics in the presence of non-Gaussian noise by using CM data). This approach uses multiple dynamic system models to explain different degradation stages, utilizing robust Bayesian estimation. As this approach is based on dynamic behavior, a threshold for diagnostics is no longer needed. Ultimately, the proposed approach is applied to the online diagnosis of simulated and actual data sets. The results of both simulated and real data sets prove the method’s efficacy.
... The number of identified levels and the methods used for their identification vary and sophistication raises compared to the two-stage case. From using confidence levels to build a four-stage system [49], to analyse changes of frequency amplitudes in the power spectral density to develop a five-stage model [50], and apply classification or clustering algorithms, such as K-nearest neighbour [51], fuzzy cmeans [52] and K-means [53], to develop the multi-stage division of machinery, the available methodologies for researchers are numerous. For those industrial-oriented real applications though, semi-supervised learning has recently emerged as a prominent methodology [54] [55] [56], as in practical applications, obtaining accurate labels based on real-time bearing conditions can be far more challenging and semi-supervised approach allows for effective utilization of dataset when only a small subset of data have labels. ...
Condition monitoring together with predictive maintenance of bearings and other equipment used by the industry avoids severe economic losses resulting from unexpected failures, greatly improves the system reliability and allows a more efficient usage of human experts' time. This paper describes a predictive maintenance system, based on a data science approach. The system was developed and tested on a single real paper machine, and then verified with multiple external validations. Results show a proper behavior of the approach on predicting different machine states with high accuracy.
... Therefore, the efficient process monitoring of thermal power Manuscript plants is essential. Among monitoring algorithms [2], [3], multivariate statistical methods have gradually become a research hotspot among scholars because they are good at mining big data information and require less prior knowledge [4], [5], [6]. Classic multivariate statistical algorithms include principal component analysis (PCA) [7], [8], [9], partial least squares, and independent component analysis (ICA). ...
Accurate process monitoring plays a crucial role in thermal power plants since it constitutes large-scale industrial equipment and its production safety is of great significance. Therefore, accurate process monitoring is very important for thermal power plants. The vigorous nature of the production process requires dynamic algorithms for monitoring. Since the common dynamic algorithm is mainly based on data expansion, the online computing complexity is too high because of data redundancy. Accordingly, this article proposes an innovative, dynamic process monitoring algorithm called autocorrelation feature analysis (AFA). AFA mines the dynamic information of continuous samples by calculating the correlation between the current time and past time features. While improving the monitoring effect, the AFA algorithm also has extremely low online computational complexity, even lower than common static algorithms, such as principal component analysis. Furthermore, this study exhibits the general form of dynamic additive faults for the first time and verifies the reliability of the algorithm through fault detectability analysis. Conclusively, the superiority of the AFA algorithm is verified on a numerical example, continuous stirred tank reactor (CSTR), and real data measured from a 1000-MW ultrasupercritical thermal power plant.
