Figure - available from: Neural Computing and Applications
This content is subject to copyright. Terms and conditions apply.
Source publication
The time variance and nonlinearity of forging processes pose great challenges to high-quality production. In this study, a one-step-ahead model predictive control (MPC) strategy based on backpropagation (BP) neural network is proposed for the precise forging processes. Two online updated BP neural networks, predictive neural network (PNN) and contr...
Similar publications
Streamflow forecasting is essential for hydrological engineering. In accordance with the advancement of computer aids in this field, various machine learning (ML) models have been explored to solve this highly non-stationary, stochastic, and nonlinear problem. In the current research, a newly explored version of an ML model called the long short-te...
Scholars have paid considerable attention to the factors that affect the safety states of construction workers. However, only a few studies have focused on the safety assessment and security alerts of individual workers. In this study, the term ‘frequency statistics’ refers to the factors considered by domestic and foreign experts and scholars. The...
Federated learning (FL) is increasingly becoming the default approach for training machine learning models across decentralized Internet-of-Things (IoT) devices. A key advantage of FL is that no raw data are communicated across the network, providing an immediate layer of privacy. Despite this, recent works have demonstrated that data reconstructio...
Historical wind data analysis is a key part in estimating design wind loads. Current design standards do not separately consider the wind loading effects by different wind types. One reason for this lack of consideration is that the segmentation between thunderstorm and non-thunderstorm data is still an issue to obtain separate data for analysis. A...
Pencurian kendaraan di Jawa Timur, merupakan salah satu bentuk kejahatan yang menimbulkan keresahan dan kecemasan masyarakat. Peningkatan tingkat pencurian kendaraan di sebabkan oleh beberapa faktor, antara lain faktor ekonomi, faktor lingkungan, rendahnya pendidikan, meningkatnya pengangguran dan kurangnya kesadaran hukum. Salah satu upaya untuk m...
Citations
... Nowadays, the intelligent optimization algorithms and machine learning model have been applied in the complex optimization problems [22][23][24] and model predicted control [25][26][27], such as simulated annealing algorithm, particle swarm optimization (PSO) algorithm, and back propagation neural networks (BPNN) model, etc. Based on the simulated annealing algorithm, Ren et al. [28] developed a hierarchical optimization method for 3D integrated circuits. ...
In this research, based on the back propagation neural network (BPNN) model and particle swarm optimization with linear decreasing inertia weight (PSO-LDIW) algorithm, an electro-thermal and thermal-stress fields coupling optimization design method for coaxial through silicon via (CTSV) is provided. The irregular and complex relationship between the parameters (height of CTSV, radius, thickness of SiO2, BCB and coaxial annular) and indexes is investigated by COMSOL Multiphysics and HFSS software. According to the simulation data of COMSOL and HFSS, the BPNN models are used to express the corresponding relationship between the parameters and indexes of CTSV. According to the desired indexes of CTSV, the multi-objective optimization function is formulated. Then, the PSO-LDIW algorithm is applied to optimize the parameters of CTSV. Finally, the simulation experiment is used to verify the effectiveness of the optimization design strategy. The simulated indexes (-40.28 dB, 366.31 K, 115.21, 74.14 and 23.16 MPa) well consent to the desired ones (-40 dB, 360 K, 110, 70 and 25 MPa), which indicates that the parameters of CTSV can effectively optimized by the developed design method to control the indexes. Therefore, the developed electro-thermal and thermal-stress fields coupling optimization design method can effectively design CTSV for manufacturing high-performance Chiplet-based microsystem.
... Consequently, the semi-closed loop is beneficial for enhancing the accuracy of the crank position control, particularly when clearance is present. A one-step MPC strategy based on a backpropagation neural network is proposed for the precise forging processes [20]. Another study derives a dynamic model by the Lagrange approach for a servo crank press machine [21]. ...
Nonlinear factors in industrial servo press currently, including mechanisms for force amplification, installed clearances, gears backlash, and variations in rotational inertia, are commonly not considered during the control process. They would impact the control accuracy of the actuator's crank and slide and even affect the quality of material forming. So the conventional linear control method, field-oriented control based on the mini-closed loop of the electrical machine position utilizing kinematics, can not follow the trajectory smoothly and accurately. This study proposes a solution to address these issues by introducing a semi-closed loop control approach. This method fully accounts for the Lagrangian approach for the dynamic model of the servo press. The torque reference of the computed torque algorithm is implemented by model predictive current control. A comprehensive and comparative trial and analysis of the impacts of critical parameters are conducted. The proposed method is practically tested on a 1600 kN press equipped with a dual-core DSP servo drive. Experimental results demonstrate the effectiveness of this control approach, which significantly enhances the control accuracy of crank position and speed while also reducing the switching frequency and losses of the servo press drive.
... Therefore, the tools are exposed to the operation of many destructive factors, which cause their wear, both premature and after a longer operation time. The areas especially subjected to wear are the working surface and the surface area of the tool, and so, most of the mentioned destructive mechanisms refer to those areas of the tool [11,12]. Exploitation is an indispensable phenomenon accompanying the production of products and is most often associated with the maximum use of a machine/system/tool in a specific time, after which it ends with their partial or complete wear. ...
... Exploitation is an indispensable phenomenon accompanying the production of products and is most often associated with the maximum use of a machine/system/tool in a specific time, after which it ends with their partial or complete wear. Therefore, in the technical literature, a lot of space is devoted to the issue of exploitation and a number of studies have been carried out on the determination of significant parameters affecting this phenomenon, as well as industrial research and development works that allow researchers to increase the exploitation time [11][12][13]. The most commonly occurring and main destructive mechanisms include abrasive wear [14] and plastic deformation, during warm forging [15][16][17], as well as plastic deformations in the case of hot forging [18,19]. ...
The paper presents the results of tests on a die insert made of non-standardised chrome-molybdenum–vanadium tool steel used during pre-forging, the life of which was 6000 forgings, while the average life for such tools is 8000 forgings. It was withdrawn from production due to intensive wear and premature breakage. In order to determine the causes of increased tool wear, a comprehensive analysis was carried out, including 3D scanning of the working surface; numerical simulations, with particular emphasis on cracking (according to the C-L criterion); and fractographic and microstructural tests. The results of numerical modelling in conjunction with the obtained results of structural tests allowed us to determine the causes of cracks in the working area of the die, which were caused by high cyclical thermal and mechanical loads and abrasive wear due to intensive flow of the forging material. It was found that the resulting fracture initiated as a multi-centric fatigue fracture continued to develop as a multifaceted brittle fracture with numerous secondary faults. Microscopic examinations allowed us to evaluate the wear mechanisms of the insert, which included plastic deformation and abrasive wear, as well as thermo-mechanical fatigue. As part of the work carried out, directions for further research were also proposed to improve the durability of the tested tool. In addition, the observed high tendency to cracking of the tool material used, based on impact tests and determination of the K1C fracture toughness factor, led to the proposal of an alternative material characterised by higher impact strength.
... Since the BPNN algorithm applies the gradient descent and is suitable for the nonlinear mapping, it is a common algorithm and is extensively applied to predict in different situations [18][19][20]. Lin et al. [21] illustrated an online prediction model of control strategy for forging hydraulic machines by BPNN. Li et al. [22] proposed an approach to predict heterogeneous network link. ...
Two data-driven algorithms, back propagation neural network (BPNN) and support vector regression (SVR), are adopted to predict air quality index (AQI) in Jiangsu Province. Meanwhile, the static model, the dynamic model of daily training and the half-daily training are established to validate the performance of algorithms comprehensively. The fundamental advantage of support vector is that less data in the full set is selected to efficiently capture the whole characteristics, whereas BPNN is more accurate in description of the high dimensional models since the parameters are well trained. The comparisons between two algorithms for the above models demonstrate that BPNN outperforms SVR in terms of accuracy since most of the mean absolute percentage errors of BPNN are less than 10%, which decrease 3% compared with that of SVR, whereas the computational cost of SVR is much less than that of BPNN. Furthermore, a novel hybrid model, the SVR-BPNN model, is proposed to further predict and analyze the AQI, which performs as fairly well as BPNN but is less time-consuming.
... The second kind of model is established according to the deformation mechanism of the material, and the microscopic aspects, such as dislocation accumulation and grain size evolution, are considered [30][31][32]. The third kind of constitutive model can obtain the material constants using regression analysis and is used by many researchers due to its high prediction accuracy [33][34][35]. ...
Hastelloy C276 is widely used in the new generation of nuclear power plants, and hot deformation is the optimum way to form the C276 part. In this investigation, the hot deformation and constitutive modeling of Hastelloy C276 alloy are researched, and the processing maps are drawn. The results show that strain rate and hot deformation temperature have remarkable impacts on the deformation behaviors of the Hastelloy C276 alloy. The yield behavior and the flow stress are predicted based on the Arrhenius constitutive equation, and the correlation coefficients are 0.9613 and 0.9837, indicating the high prediction ability of the established constitutive equation. Rising the temperature can decrease the unstable deformation area, and the studied alloy can be deformed at low strain rates. With the increased strain rate, flow localization occurs, which is not suitable for hot deformation.
... Recently, the artificial neural network (ANN), has been widely used in the model predicted control [15][16][17], image processing [18][19][20] and optimization algorithms [21][22][23]. In addition, the particle swarm optimization (PSO) algorithm has been utilized in the topology optimization for compliant mechanisms [24,25], the optimal design parameters of electronic components [26] and the optimal parameters of the filtering algorithm [27,28]. ...
In this paper, a thermal-stress coupling optimization strategy for coaxial through silicon via (TSV) is developed based on the finite element method (FEM), artificial neural network (ANN) model and particle swarm optimization (PSO) algorithm. In order to analyze the effect of design parameters on the thermal-stress distribution of coaxial TSV, the FEM simulations of coaxial TSV are conducted by COMSOL Multiphysics. The structure of coaxial TSV is symmetric. The mapping relationships between the design parameters and performance indexes are described by ANN models based on the simulation data of FEM. In addition, the multi-objective optimization function is formulated based on the desired performance indexes, and then the design parameters are optimized by the modified PSO algorithm. Based on the optimized design parameters, the effectiveness of the developed method is validated by FEM simulations. The simulated performance indexes agree well with the desired ones, which implies that the design parameters of coaxial TSV can be optimized to control the thermal-stress distribution. Therefore, the thermal-stress coupling optimization of coaxial TSV can achieve thermal-stress management to improve its reliability.
... However, most industrial objects are nonlinear, and the effectiveness and robustness of many predictive control methods for the realtime optimization are deficient, making many successful linear system models poorly exploited. A large number of MPCs, based on neural networks, were subsequently investigated, such as the ant colony optimization (ACO) [20], gradient descent (GD) [21], genetic algorithm (GA) [22], support vector machine (SVM) [23], back propagation (BP) [24] and radial basis function (RBF) [25]. However, the learning efficiency and the generalization ability of these predictive control algorithms were far below the industrial requirements. ...
Raise boring is an important method to construct the underground shafts of mines and other underground infrastructures, by drilling down the pilot hole and then reaming up to the desired diameter. Seriously different from the drilling operations of the mechanical parts in mechanized mass production, it is very difficult to obtain a good consistency in the construction environments of each raise or shaft, to be more exact, every construction process is highly customized. The underground bottom-up reaming process is impossible to be observed directly, and the rock breaking effect is very difficult to be measured in real-time, due to the rock debris freely falling under the excavated shaft. The optimal configurations of the operational parameters in the drilling and working pressures, torque, rotation speed and penetration speed, mainly depend on the accumulation of construction experience or empirical models. To this end, we presented a machine learning method, based on the extreme learning machine, to determine in real-time, the relationships between the working performance and the operational parameters, and the physical-mechanical properties of excavated geologic zones, aiming at a higher production or excavation rate, safer operation and minimum ground disturbance. This research brings out new possibilities to revolutionize the process planning paradigm of the raise boring method that traditionally depends on experience or subject matter expertise.
... Different from traditional mathematic models, the data-driven ML model is like a black box that can easily handle the complex physical relationship among variables and considerate configuration parameters and environmental variables in the model [23], [24]. Currently, the relative application of ML to hydraulic systems mainly concentrates on fault diagnosis [25], [26], monitoring [27], and nonlinear control [28] (e.g., neural network control [29], [30], [31]), where uncertainty parameters and unknown disturbances are successfully handled. Therefore, an ML-based energy model is a promising method to achieve the ideal energysaving mode. ...
In this paper, an artificial intelligence enabled energy-saving strategy (AIESS) is proposed to achieve high energy efficiency of hydraulic systems with a speed and displacement variable pump (SDVP). The AIESS, which integrates extreme gradient boosting (XGBoost) and genetic algorithm (GA), can generate the optimum combination of motor speed and pump displacement of SDVP in real time under actual working conditions. The XGBoost-based power model is used to accurately predict the input power of the SDVP considering environmental factors, e.g., oil temperature, then followed by GA for optimization, in which the XGBoost-based power model is the objective function. A test rig with a 16t hydraulic press and a stamping process is applied to validate the effectiveness of the proposed strategy. Compared to four other algorithm models, the XGBoost-based power model has the highest accuracy. The optimum combination can be obtained quickly by GA. The results show that the AIESS can reduce energy consumption by 28.6% during pressure-maintaining operation and by 11.2% during one stamping cycle compared to the previously proposed segmented control energy-saving strategy.
Note to Practitioners
—Energy-saving hydraulic system is a topic in the automation industry. The low energy efficiency of electro-hydraulic systems is mainly caused by poor flow rate matching. A speed and displacement variable pump (SDVP) is used to address this problem, but it is difficult to find the matched speed and displacement for the highest energy conversion efficiency of the SDVP due to the inaccuracy energy model and the challenge of coupling variables. This paper presents an energy-saving strategy integrated with machine learning and a genetic algorithm to handle this challenge. Relative variables of the machine learning model are picked through analysis of the mathematical energy model, and the usage of machine learning can fit the high-nonlinear energy model caused by uncertainties in both components and environmental factors. The trained model is then used as an objective function for a GA-based optimization process to generate the optimum motor speed and pump displacement combination. Considering the processing time of GA, this combination can be optimized before the operation for known working conditions. Finally, we have established a 16t hydraulic press to validate this energy-saving strategy and pre-optimized five combinations for a stamping process. The proposed strategy can reduce energy consumption by 11.2% compared with the previous strategy. This energy-saving strategy can also be applied to other automated hydraulic equipment, such as injection machines.
... All networks contain an input layer and an output layer. e main difference is that the number of hidden layers in the middle is different [22]. e hidden layer is not directly connected to the outside world, but its changes will have a direct impact on the relationship between the input layer and the output layer [23]. ...
The objectives are to solve many problems in traditional English reading teaching, such as the passive acceptance of students’ learning situation, the rigid teaching mode of teachers and the difficulty in taking into account the individual needs of each student, and the forced averaging of students’ English learning ability and level. Firstly, a detection method of English reading level based on the Backpropagation neural network (BPNN) is designed. Secondly, the teaching process and specific teaching plan are designed with the Production Oriented Approach (POA). Finally, the detection method of English reading level is verified by simulation experiments, and the effect of the teaching method integrating POA is analyzed. The results show that (1) when the learning rate is 0.26 and the training time is 1000 times, the mean square error (MSE) of the model is the smallest. (2) The results of the practical experiment indicate that the number of people who sign up for the college English proficiency test, the number of actual exams, and the pass rate of the test in the experimental group using the teaching method are higher than those in the control group. Moreover, the number of people in the experimental group increased more than that in the control group after the experiment. There is no significant difference in the test results of the two groups before the experiment, and the average test scores of the experimental group are obviously higher than those of the control group after the experiment. The p value of the two groups of results is less than 0.05 in the t-test, indicating that there is a significant difference in the English reading level between the two groups after the experiment. Simultaneously, the difference between the results of class A before and after the experiment is larger than that of class B, showing that the teaching method integrating POA is more effective for students with higher English levels. Therefore, compared with the traditional teaching method, the teaching mode integrating POA is more effective in improving the English reading level of college students. It aims to transform POA theory into practice and provide a great reference for college English reading teaching.
... The recent studies on Artificial Intelligence with increased computing capability attract attention of researchers to MPC enhanced by AI learning methods. Lin et al. [13] used BP neural network-based online model predictive control strategy to deal with the time variance and nonlinearity of forging process. Overcome nonlinearities and low output accuracy of the transplanting manipulator. ...
