Wei Huang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Wei Huang has a total of 28 co-authors. Here are the top co-authors by number of shared publications:
Jian Xu
China National Machinery Industry Corporation Ltd (SINOMACH) Beijing China
11 shared publications- Jian Xu
Sinomach (China)
3 shared publications Xu Jian
China National Machinery Industry Corporation Ltd (SINOMACH), Beijing, China
3 shared publicationsHu Ming-Yi
2 shared publicationsZhang Tong-Yi
2 shared publicationsCheng Liu
School of Civil and Hydraulic Engineering, Hefei University of Technology, Hefei, China and Anhui Provincial Laboratory of Civil Engineering and Materials, Hefei, China
1 shared publication
Latest Publications
- Open Access
- 21 May 2026
- Published by Wiley in AI & Innovation
- Vol. 1 (1)
PPO‐Based Reinforcement Learning for the Semi‐Active Vibration Control of MDOF Platform
Aiming at the coupled vibration problem of a multi‐degree‐of‐freedom (MDOF) vibration isolation platform under eccentric excitation, this paper proposes a semi‐active vibration control strategy based on Proximal Policy Optimization (PPO) ‐based reinforcement learning (PPO RL). First, the mechanical model of a 3‐DOF platform supported by four magnetorheological dampers (MRDs) and isolators is established. On this basis, the semi‐active vibration control problem is formulated as a Markov Decision Process (MDP). A 20‐dimensional state space is constructed, integrating the motion of the centroid, local support point motion, vibration excitation, and MRD internal hysteresis. Accordingly, a 4‐dimensional continuous action space corresponding to the input current of each MRD is designed. A multi‐objective reward function is developed, which accounts for vibration suppression, differential vibration penalty and control input efficiency. The actor‐critic dual network of the PPO algorithm is optimized with carefully tailored hyperparameters, and GPU acceleration is employed to enhance training efficiency. Simulation results demonstrate that the proposed strategy significantly suppresses both the vertical translation and the angular vibration of the platform's centroid, while effectively controlling the differential vibration of support points. More importantly, by suppressing the angular vibration of the centroid and ensuring the consistency of support point displacement, the strategy achieves high‐precision horizontal attitude maintenance of the platform under eccentric excitation. The maximum relative deviation of the support point displacement in z ‐direction is reduced from 28.5% to 6.2%. This research verifies the effectiveness and superiority of the PPO algorithm for semi‐active vibration control of MDOF vibration isolation platforms, and provides a new intelligent control method for the attitude and vibration control of isolation platforms under asymmetric excitation.
- 21 May 2026
- Published by Wiley in AI & Innovation
- Vol. 1 (1)
Open AccessPPO‐Based Reinforcement Learning for the Semi‐Active Vibration Control of MDOF Platform
Aiming at the coupled vibration problem of a multi‐degree‐of‐freedom (MDOF) vibration isolation platform under eccentric excitation, this paper proposes a semi‐active vibration control strategy based on Proximal Policy Optimization (PPO) ‐based reinforcement learning (PPO RL). First, the mechanical model of a 3‐DOF platform supported by four magnetorheological dampers (MRDs) and isolators is established. On this basis, the semi‐active vibration control problem is formulated as a Markov Decision Process (MDP). A 20‐dimensional state space is constructed, integrating the motion of the centroid, local support point motion, vibration excitation, and MRD internal hysteresis. Accordingly, a 4‐dimensional continuous action space corresponding to the input current of each MRD is designed. A multi‐objective reward function is developed, which accounts for vibration suppression, differential vibration penalty and control input efficiency. The actor‐critic dual network of the PPO algorithm is optimized with carefully tailored hyperparameters, and GPU acceleration is employed to enhance training efficiency. Simulation results demonstrate that the proposed strategy significantly suppresses both the vertical translation and the angular vibration of the platform's centroid, while effectively controlling the differential vibration of support points. More importantly, by suppressing the angular vibration of the centroid and ensuring the consistency of support point displacement, the strategy achieves high‐precision horizontal attitude maintenance of the platform under eccentric excitation. The maximum relative deviation of the support point displacement in z ‐direction is reduced from 28.5% to 6.2%. This research verifies the effectiveness and superiority of the PPO algorithm for semi‐active vibration control of MDOF vibration isolation platforms, and provides a new intelligent control method for the attitude and vibration control of isolation platforms under asymmetric excitation.
- Open Access
- 19 May 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Parallel fusion model for complex multi-source vibration time-series prediction
Aiming at the insufficient prediction accuracy caused by the non-stationary and multi-frequency coupling characteristics of complex multi-source vibration signals, this study proposes a parallel fusion prediction model that takes hyperparameter optimization as the core preprocessing step and integrates multivariate variational mode decomposition (MVMD) with parallel weighted KNN-XGBoost. Firstly, a "training-validation-test" three-set data separation framework is constructed, and a stepwise grid search strategy is adopted to optimize key hyperparameters. With the error validation set as the objective function, the number of neighbors for K-Nearest Neighbors (KNN), the tree depth and learning rate for eXtreme Gradient Boosting (XGBoost), and the optimal fusion weights of KNN and XGBoost are determined sequentially. This fundamentally avoids the subjective bias inherent in manual parameter tuning. Secondly, MVMD is employed to decompose complex multi-source vibration signals into 8 intrinsic mode functions (IMFs) with different frequency scales. This addresses the mode mixing issue of traditional decomposition methods, reduces the complexity of vibration signals, and retains the multi-source coupling features. Thirdly, a parallel weighted KNN-XGBoost structure is built based on the optimized parameters. Through independent prediction and weighted fusion, error propagation in serial structures is avoided, enabling the synergy of "local correction-global fitting". Multiple comparative and ablation experiments demonstrate that, compared with other models, the MVMD-parallel-KNN-XGBoost model achieves the optimal balanced performance. This study provides a parallel fusion technical pathway of "parameter optimization-modal separation-model fusion" for complex multi-source vibration time-series prediction, which can meet the high-precision prediction requirements of complex multi-source vibration in fields such as aerospace, rail transit, and industrial engineering etc.
- 19 May 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessParallel fusion model for complex multi-source vibration time-series prediction
Aiming at the insufficient prediction accuracy caused by the non-stationary and multi-frequency coupling characteristics of complex multi-source vibration signals, this study proposes a parallel fusion prediction model that takes hyperparameter optimization as the core preprocessing step and integrates multivariate variational mode decomposition (MVMD) with parallel weighted KNN-XGBoost. Firstly, a "training-validation-test" three-set data separation framework is constructed, and a stepwise grid search strategy is adopted to optimize key hyperparameters. With the error validation set as the objective function, the number of neighbors for K-Nearest Neighbors (KNN), the tree depth and learning rate for eXtreme Gradient Boosting (XGBoost), and the optimal fusion weights of KNN and XGBoost are determined sequentially. This fundamentally avoids the subjective bias inherent in manual parameter tuning. Secondly, MVMD is employed to decompose complex multi-source vibration signals into 8 intrinsic mode functions (IMFs) with different frequency scales. This addresses the mode mixing issue of traditional decomposition methods, reduces the complexity of vibration signals, and retains the multi-source coupling features. Thirdly, a parallel weighted KNN-XGBoost structure is built based on the optimized parameters. Through independent prediction and weighted fusion, error propagation in serial structures is avoided, enabling the synergy of "local correction-global fitting". Multiple comparative and ablation experiments demonstrate that, compared with other models, the MVMD-parallel-KNN-XGBoost model achieves the optimal balanced performance. This study provides a parallel fusion technical pathway of "parameter optimization-modal separation-model fusion" for complex multi-source vibration time-series prediction, which can meet the high-precision prediction requirements of complex multi-source vibration in fields such as aerospace, rail transit, and industrial engineering etc.
- Open Access
- 1 March 2026
- Published by KEAI Publishing in Journal of Dynamic Disasters
- Vol. 2 (1)
Study on indoor secondary radiated noise under integrated control of engineering vibration and seismic vibration of building engineering
- 1 March 2026
- Published by KEAI Publishing in Journal of Dynamic Disasters
- Vol. 2 (1)
Open AccessStudy on indoor secondary radiated noise under integrated control of engineering vibration and seismic vibration of building engineering
- 18 February 2026
- Published by Springer Nature in Journal of Vibration Engineering & Technologies
- Vol. 14 (3) , 109
Engineering Vibration Recognition Using CWT-CNN
- 18 February 2026
- Published by Springer Nature in Journal of Vibration Engineering & Technologies
- Vol. 14 (3) , 109
Engineering Vibration Recognition Using CWT-CNN
- Open Access
- 8 January 2026
- Published by Wiley in Advances in Civil Engineering
- Vol. 2026 (1)
Prediction of Nuclear Explosion Vibration and MRD Semi‐Active Control Based on Deep Learning
This research centers on the application of deep learning in predicting nuclear explosion vibration loads and the semi‐active control of magnetorheological dampers (MRDs). It delves deeply into the performance of the Convolutional Neural Network (CNN) and the long short‐term memory (LSTM) network. The time history of impact acceleration in the near and far zones of a nuclear explosion was simulated, and the prediction performances of the two models under different ratios of the training set to the test set were compared. The results indicate that the LSTM has a faster training speed and performs better in most training set error indices. In terms of the test set, the two models have their respective advantages and disadvantages under different ratios. Regarding the prediction of the semi‐active control effect of MRDs, a corresponding system was constructed, and calculations and analyses were carried out. It was found that the LSTM has high training efficiency and generally better performance on the training set. On the test set, the LSTM performs slightly better when the ratio is 0.8:0.2, while the CNN performs better when the ratio is 0.5:0.5. This study provides a theoretical basis and technical support for nuclear explosion vibration prediction and the improvement of the blast resistance of protective structures. It also points out that future research can further optimize the model structure and parameters and explore more effective data processing methods to expand the application of deep learning models in the fields of military and protective engineering.
- 8 January 2026
- Published by Wiley in Advances in Civil Engineering
- Vol. 2026 (1)
Open AccessPrediction of Nuclear Explosion Vibration and MRD Semi‐Active Control Based on Deep Learning
This research centers on the application of deep learning in predicting nuclear explosion vibration loads and the semi‐active control of magnetorheological dampers (MRDs). It delves deeply into the performance of the Convolutional Neural Network (CNN) and the long short‐term memory (LSTM) network. The time history of impact acceleration in the near and far zones of a nuclear explosion was simulated, and the prediction performances of the two models under different ratios of the training set to the test set were compared. The results indicate that the LSTM has a faster training speed and performs better in most training set error indices. In terms of the test set, the two models have their respective advantages and disadvantages under different ratios. Regarding the prediction of the semi‐active control effect of MRDs, a corresponding system was constructed, and calculations and analyses were carried out. It was found that the LSTM has high training efficiency and generally better performance on the training set. On the test set, the LSTM performs slightly better when the ratio is 0.8:0.2, while the CNN performs better when the ratio is 0.5:0.5. This study provides a theoretical basis and technical support for nuclear explosion vibration prediction and the improvement of the blast resistance of protective structures. It also points out that future research can further optimize the model structure and parameters and explore more effective data processing methods to expand the application of deep learning models in the fields of military and protective engineering.
- 27 March 2025
- Published by SAGE Publications in Noise & Vibration Worldwide
- Vol. 56 (4-5) , 224-241
Application of phononic crystals in modern engineering vibration and noise control, a review
Phononic crystals are composite materials consisting of periodically distributed inclusions, and this structure enables wave band gaps under specific conditions. This characteristic allows for the suppression of elastic waves within certain frequency ranges, offering a new approach to vibration reduction and noise control. In this study, phononic crystals, with their periodic structure, tunable bandgap properties, topology optimization, and additive manufacturing technology, are reviewed. Some key hot issues, such as low-frequency vibration and noise control, intelligent materials, and tunable bandgap design, are discussed. In the future, phononic crystals will play a significant role in barrier vibration isolation, rail transit vibration isolation, power equipment vibration isolation, precision equipment vibration isolation, noise control, and underground protection engineering. This study constitutes a systematic summary and outlook of phononic crystals in modern engineering vibration and noise control.
- 27 March 2025
- Published by SAGE Publications in Noise & Vibration Worldwide
- Vol. 56 (4-5) , 224-241
Application of phononic crystals in modern engineering vibration and noise control, a review
Phononic crystals are composite materials consisting of periodically distributed inclusions, and this structure enables wave band gaps under specific conditions. This characteristic allows for the suppression of elastic waves within certain frequency ranges, offering a new approach to vibration reduction and noise control. In this study, phononic crystals, with their periodic structure, tunable bandgap properties, topology optimization, and additive manufacturing technology, are reviewed. Some key hot issues, such as low-frequency vibration and noise control, intelligent materials, and tunable bandgap design, are discussed. In the future, phononic crystals will play a significant role in barrier vibration isolation, rail transit vibration isolation, power equipment vibration isolation, precision equipment vibration isolation, noise control, and underground protection engineering. This study constitutes a systematic summary and outlook of phononic crystals in modern engineering vibration and noise control.
- Open Access
- 7 January 2025
- Published by Academic Publishing Pte. Ltd. in Sound & Vibration
- Vol. 59 (1) , 2242
Engineering vibration recognition using CWT-ResNet
Multi-source signal recognition is a common problem in engineering vibration control. Given that traditional methods often primarily rely on prior knowledge and expertise, which can limit efficiency and accuracy, this study proposed a vibration recognition model based on ResNet, utilizing continuous wavelet transform to combine signal processing with deep learning techniques. The continuous wavelet transform converts the original one-dimensional vibration signals into two-dimensional time-frequency representations with richer feature information, which are then input into the convolutional layers for automatic feature extraction, culminating in vibration recognition through the Softmax layer. To evaluate the model’s performance, 20 sets of measured vibration data were tested. The results show that the proposed model achieves a recognition accuracy of 99%, excelling in both component recognition and the separation of vibration signals. Therefore, this study is of great significance for engineering vibration diagnosis, the front-end design of vibration control, and the analysis and optimization of control effectiveness.
- 7 January 2025
- Published by Academic Publishing Pte. Ltd. in Sound & Vibration
- Vol. 59 (1) , 2242
Open AccessEngineering vibration recognition using CWT-ResNet
Multi-source signal recognition is a common problem in engineering vibration control. Given that traditional methods often primarily rely on prior knowledge and expertise, which can limit efficiency and accuracy, this study proposed a vibration recognition model based on ResNet, utilizing continuous wavelet transform to combine signal processing with deep learning techniques. The continuous wavelet transform converts the original one-dimensional vibration signals into two-dimensional time-frequency representations with richer feature information, which are then input into the convolutional layers for automatic feature extraction, culminating in vibration recognition through the Softmax layer. To evaluate the model’s performance, 20 sets of measured vibration data were tested. The results show that the proposed model achieves a recognition accuracy of 99%, excelling in both component recognition and the separation of vibration signals. Therefore, this study is of great significance for engineering vibration diagnosis, the front-end design of vibration control, and the analysis and optimization of control effectiveness.
- Open Access
- 1 January 2025
- Published by Wiley in Shock and Vibration
- Vol. 2025 (1)
Engineering Vibration Recognition Using CWT‐VGG19
Multisource signal recognition is a common problem in engineering vibration control. Given that traditional methods often primarily rely on prior knowledge and expertise, which can limit efficiency and accuracy, this study proposed a vibration recognition model based on VGG19, utilizing continuous wavelet transform to combine signal processing with deep learning techniques. The continuous wavelet transform converts the original one‐dimensional vibration signals into two‐dimensional time‐frequency representations with richer feature information, which are then input into the convolutional layers for automatic feature extraction, culminating in vibration recognition through the SoftMax layer. To evaluate the model’s performance, 20 sets of measured vibration data were tested. The results show that the proposed model achieves a recognition accuracy of 99%, excelling in both component recognition and the separation of vibration signals. Therefore, this study is of great significance for engineering vibration diagnosis, the front‐end design of vibration control, and the analysis and optimization of control effectiveness.
- 1 January 2025
- Published by Wiley in Shock and Vibration
- Vol. 2025 (1)
Open AccessEngineering Vibration Recognition Using CWT‐VGG19
Multisource signal recognition is a common problem in engineering vibration control. Given that traditional methods often primarily rely on prior knowledge and expertise, which can limit efficiency and accuracy, this study proposed a vibration recognition model based on VGG19, utilizing continuous wavelet transform to combine signal processing with deep learning techniques. The continuous wavelet transform converts the original one‐dimensional vibration signals into two‐dimensional time‐frequency representations with richer feature information, which are then input into the convolutional layers for automatic feature extraction, culminating in vibration recognition through the SoftMax layer. To evaluate the model’s performance, 20 sets of measured vibration data were tested. The results show that the proposed model achieves a recognition accuracy of 99%, excelling in both component recognition and the separation of vibration signals. Therefore, this study is of great significance for engineering vibration diagnosis, the front‐end design of vibration control, and the analysis and optimization of control effectiveness.
- 28 September 2024
- Published by SAGE Publications in Noise & Vibration Worldwide
- Vol. 55 (11) , 632-650
Magnetorheological damper parameter identification and its application in semi-active control of power equipment
In recent years, magnetorheological dampers (MRD) have played a significant role in vibration control in various fields such as vehicles, military, building structures, etc. However, the mechanical model of MRD is very complex due to its hysteresis, which makes it difficult to identify the model’s parameters. In addition to experimental data, under the condition of no other prior knowledge, eight unknown parameters of the Bouc-Wen model are identified in this study based on the particle swarm optimization (PSO) algorithm. According to the trend of parameter variation with current, fitting research is conducted, and the variation law of parameters with current is summarized. Based on the identified parameters, the MRD’s output force is predicted under any current other than the experimental current. Subsequently, the application of MRD in semi-active control of power equipment is performed out, and proposed Simulink calculation programs for single-stage and two-stage vibration control systems are built. Then, a comparative study with passive and active control is conducted. In this paper, linear quadratic regulator (LQR) control is adopted for the active control, and the controller’s parameters are optimized based on the PSO algorithm. The results show that the adopted semi-active control can significantly reduce the transmitted force from power equipment to the foundation and can effectively mitigate the disturbance of power equipment to the environment. This study offers important guidance for the vibration control of MRD in industrial engineering.
- 28 September 2024
- Published by SAGE Publications in Noise & Vibration Worldwide
- Vol. 55 (11) , 632-650
Magnetorheological damper parameter identification and its application in semi-active control of power equipment
In recent years, magnetorheological dampers (MRD) have played a significant role in vibration control in various fields such as vehicles, military, building structures, etc. However, the mechanical model of MRD is very complex due to its hysteresis, which makes it difficult to identify the model’s parameters. In addition to experimental data, under the condition of no other prior knowledge, eight unknown parameters of the Bouc-Wen model are identified in this study based on the particle swarm optimization (PSO) algorithm. According to the trend of parameter variation with current, fitting research is conducted, and the variation law of parameters with current is summarized. Based on the identified parameters, the MRD’s output force is predicted under any current other than the experimental current. Subsequently, the application of MRD in semi-active control of power equipment is performed out, and proposed Simulink calculation programs for single-stage and two-stage vibration control systems are built. Then, a comparative study with passive and active control is conducted. In this paper, linear quadratic regulator (LQR) control is adopted for the active control, and the controller’s parameters are optimized based on the PSO algorithm. The results show that the adopted semi-active control can significantly reduce the transmitted force from power equipment to the foundation and can effectively mitigate the disturbance of power equipment to the environment. This study offers important guidance for the vibration control of MRD in industrial engineering.
- 21 May 2023
- Published by Springer Nature
- p. 173-216
Vibration Control for Equipment-Structure