Pub Date : 2023-05-12DOI: 10.1109/DDCLS58216.2023.10167264
Chiqiang Liu, Dazi Li
With the wide application of multi-intelligent reinforcement learning (MARL), its development becomes more and more mature. Multi-agent Proximal Policy Optimization (MAPPO) extended by Proximal Policy Optimization (PPO) algorithm has attracted the attention of researchers with its superior performance. However, the increase in the number of agents in multi-agent cooperation tasks leads to overfitting problems and suboptimal policies due to the fixed clip range that limits the step size of updates. In this paper, MAPPO via Non-fixed Value Clipping (NVC-MAPPO) algorithm is proposed based on MAPPO, and Gaussian noise is introduced in the value function and the clipping function, respectively, and rewriting the clipping function into a form called non-fixed value clipping function. In the end, experiments are conducted on StarCraftII Multi-Agent Challenge (SMAC) to verify that the algorithm can effectively prevent the step size from changing too much while enhancing the exploration ability of the agents, which has improved the performance compared with MAPPO.
{"title":"Multi-agent Proximal Policy Optimization via Non-fixed Value Clipping","authors":"Chiqiang Liu, Dazi Li","doi":"10.1109/DDCLS58216.2023.10167264","DOIUrl":"https://doi.org/10.1109/DDCLS58216.2023.10167264","url":null,"abstract":"With the wide application of multi-intelligent reinforcement learning (MARL), its development becomes more and more mature. Multi-agent Proximal Policy Optimization (MAPPO) extended by Proximal Policy Optimization (PPO) algorithm has attracted the attention of researchers with its superior performance. However, the increase in the number of agents in multi-agent cooperation tasks leads to overfitting problems and suboptimal policies due to the fixed clip range that limits the step size of updates. In this paper, MAPPO via Non-fixed Value Clipping (NVC-MAPPO) algorithm is proposed based on MAPPO, and Gaussian noise is introduced in the value function and the clipping function, respectively, and rewriting the clipping function into a form called non-fixed value clipping function. In the end, experiments are conducted on StarCraftII Multi-Agent Challenge (SMAC) to verify that the algorithm can effectively prevent the step size from changing too much while enhancing the exploration ability of the agents, which has improved the performance compared with MAPPO.","PeriodicalId":415532,"journal":{"name":"2023 IEEE 12th Data Driven Control and Learning Systems Conference (DDCLS)","volume":"103 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121405751","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-12DOI: 10.1109/DDCLS58216.2023.10165993
Qiu Ruikang, Li Shengquan, Cui Ronghua, Zhang Lujin, Li Juan
A linear active disturbance rejection control (LADRC) strategy is proposed to suppress the structural vibration caused by external excitations and internal uncertainties in intelligent structures under complex working conditions via an Anlu EG4S20B256 chip. First, the electromechanical coupling model of the whole vibration control system is obtained based on the dynamic equations of the all-clamped plate structure and the electromagnetic equations of the inertial actuator. Second, based on the system model, a third-order extended state observer (ESO) is designed to estimate the internal modelling errors and external excitation perturbations of the system in real time. In addition, the influence of internal and external disturbances on the control effect in the experiment is offset by a feedforward compensation. Finally, a vibration control platform based on the Anlu FPGA chip is built to verify the control effect of the proposed vibration active control strategy through physical real-time simulation.
{"title":"Intelligent Structure Control System Based on FPGA","authors":"Qiu Ruikang, Li Shengquan, Cui Ronghua, Zhang Lujin, Li Juan","doi":"10.1109/DDCLS58216.2023.10165993","DOIUrl":"https://doi.org/10.1109/DDCLS58216.2023.10165993","url":null,"abstract":"A linear active disturbance rejection control (LADRC) strategy is proposed to suppress the structural vibration caused by external excitations and internal uncertainties in intelligent structures under complex working conditions via an Anlu EG4S20B256 chip. First, the electromechanical coupling model of the whole vibration control system is obtained based on the dynamic equations of the all-clamped plate structure and the electromagnetic equations of the inertial actuator. Second, based on the system model, a third-order extended state observer (ESO) is designed to estimate the internal modelling errors and external excitation perturbations of the system in real time. In addition, the influence of internal and external disturbances on the control effect in the experiment is offset by a feedforward compensation. Finally, a vibration control platform based on the Anlu FPGA chip is built to verify the control effect of the proposed vibration active control strategy through physical real-time simulation.","PeriodicalId":415532,"journal":{"name":"2023 IEEE 12th Data Driven Control and Learning Systems Conference (DDCLS)","volume":"178 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121041175","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Facing the safety problems in industrial process, how to effectively diagnose process faults has become quite necessary and important. In this paper, a novel fault diagnosis approach integrated local reconstructed kernel principal component analysis(LRKPCA) with AdaBoost.M2 is proposed. Firstly, kernel principal component analysis(KPCA) is adopted to extract the global features through non-linear projection transformation. And local feature extraction based on t-distributed stochastic neighbor embedding(TSNE) is realized by minimizing the similarity of probability distribution of samples in high-dimensional space and low-dimensional space. Secondly, LRKPCA-based feature extraction method is proposed, in which the reconstruction error is calculated based on local features and mapped to the global feature space so that data dimension is reduced through coordinate reconstruction. Thirdly, AdaBoost.M2 is adopted to establish multi-classification model to realize fault diagnosis. Finally, the experimental results based on Tennessee Eastman process(TEP) show that the proposed method has higher diagnosis accuracy.
{"title":"A Novel Fault Diagnosis Approach Integrated LRKPCA with AdaBoost.M2 for Industrial Process","authors":"Yuan Xu, Xue Jiang, Qun Zhu, Yanlin He, Yang Zhang, Mingqing Zhang","doi":"10.1109/DDCLS58216.2023.10167144","DOIUrl":"https://doi.org/10.1109/DDCLS58216.2023.10167144","url":null,"abstract":"Facing the safety problems in industrial process, how to effectively diagnose process faults has become quite necessary and important. In this paper, a novel fault diagnosis approach integrated local reconstructed kernel principal component analysis(LRKPCA) with AdaBoost.M2 is proposed. Firstly, kernel principal component analysis(KPCA) is adopted to extract the global features through non-linear projection transformation. And local feature extraction based on t-distributed stochastic neighbor embedding(TSNE) is realized by minimizing the similarity of probability distribution of samples in high-dimensional space and low-dimensional space. Secondly, LRKPCA-based feature extraction method is proposed, in which the reconstruction error is calculated based on local features and mapped to the global feature space so that data dimension is reduced through coordinate reconstruction. Thirdly, AdaBoost.M2 is adopted to establish multi-classification model to realize fault diagnosis. Finally, the experimental results based on Tennessee Eastman process(TEP) show that the proposed method has higher diagnosis accuracy.","PeriodicalId":415532,"journal":{"name":"2023 IEEE 12th Data Driven Control and Learning Systems Conference (DDCLS)","volume":"73 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122789272","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Accurate and efficient fault root cause diagnosis is an effective means to prevent major accidents in industrial systems. Due to the difficulty of modeling complex systems, Granger causal analysis is widely used. Root cause diagnosis in the shortest possible time after a fault occurs can improve the accuracy of diagnostic results. Due to the strong nonlinear relationship in the short observation data, this paper introduces Generalized Radial Basis Function(GRBF) neural network of the nonlinear dimensionality reduction method into the Granger causal model to realize the root cause diagnosis of Granger faults based on the nonlinear short observation data. The effectiveness of the proposed method is verified by numerical simulation and fault diagnosis experimental study of Tennessee Eastman,(TE) chemical process. The results show that the proposed method improves the processing ability of Granger causal analysis for nonlinear causality, and can use a small amount of the fault data to complete accurate fault root cause diagnosis.
{"title":"A Granger causality analysis method based on GRBF network","authors":"Huang Chen, Jianguo Wang, Pangbin Ding, X. Ye, Yuan Yao, He-Lin Chen","doi":"10.1109/DDCLS58216.2023.10166901","DOIUrl":"https://doi.org/10.1109/DDCLS58216.2023.10166901","url":null,"abstract":"Accurate and efficient fault root cause diagnosis is an effective means to prevent major accidents in industrial systems. Due to the difficulty of modeling complex systems, Granger causal analysis is widely used. Root cause diagnosis in the shortest possible time after a fault occurs can improve the accuracy of diagnostic results. Due to the strong nonlinear relationship in the short observation data, this paper introduces Generalized Radial Basis Function(GRBF) neural network of the nonlinear dimensionality reduction method into the Granger causal model to realize the root cause diagnosis of Granger faults based on the nonlinear short observation data. The effectiveness of the proposed method is verified by numerical simulation and fault diagnosis experimental study of Tennessee Eastman,(TE) chemical process. The results show that the proposed method improves the processing ability of Granger causal analysis for nonlinear causality, and can use a small amount of the fault data to complete accurate fault root cause diagnosis.","PeriodicalId":415532,"journal":{"name":"2023 IEEE 12th Data Driven Control and Learning Systems Conference (DDCLS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125740544","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper develops a moving window probabilistic PCA(MW PPCA) online process monitoring method for moni-toring time-varying industrial process. First, PPCA model and the method of iteratively solving the parameters of PPCA model by variational inference are introduced. On the basis of the PPCA model, three monitoring statistic, ${T}^{2}, SPE$ and $Var$, are in-troduced also. In order to solve the time-varying trend, this paper further utilizes a sequential update procedure for PPCA model which is based on a moving window, and uses the streaming variational inference method to recursively update the parameters of PPCA model in each window. Then, the non central chi square distribution approximation is used to solve the control limits of the three statistics under the confidence limits in order to adapt to the process changes more effectively, so as to update the control limits. Finally, the effectiveness of the distillation process is verified.
{"title":"Online Monitoring of Time-varying Process Using Probabilistic Principal Component Analysis","authors":"Yuxuan Dong, Ying Liu, Suijun Liu, Cheng Lu, Shihua Luo, Jiu-sun Zeng","doi":"10.1109/DDCLS58216.2023.10166692","DOIUrl":"https://doi.org/10.1109/DDCLS58216.2023.10166692","url":null,"abstract":"This paper develops a moving window probabilistic PCA(MW PPCA) online process monitoring method for moni-toring time-varying industrial process. First, PPCA model and the method of iteratively solving the parameters of PPCA model by variational inference are introduced. On the basis of the PPCA model, three monitoring statistic, ${T}^{2}, SPE$ and $Var$, are in-troduced also. In order to solve the time-varying trend, this paper further utilizes a sequential update procedure for PPCA model which is based on a moving window, and uses the streaming variational inference method to recursively update the parameters of PPCA model in each window. Then, the non central chi square distribution approximation is used to solve the control limits of the three statistics under the confidence limits in order to adapt to the process changes more effectively, so as to update the control limits. Finally, the effectiveness of the distillation process is verified.","PeriodicalId":415532,"journal":{"name":"2023 IEEE 12th Data Driven Control and Learning Systems Conference (DDCLS)","volume":"1206 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125933271","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-12DOI: 10.1109/DDCLS58216.2023.10167248
Shiyan Li, Xuefang Li
A novel extended state observer (ESO) based iterative learning control (ILC) scheme is investigated, including three channels, namely, feedforward, feedback, and disturbance rejection channels. The goal of this work is to achieve high-accuracy tracking of nonlinear systems in the presence of nonrepetitive disturbances under repetitive operating conditions. The ESO is used to estimate and offset the nonrepetitive disturbance in real time, which reduces the sensitivity of the controller to system parameters and operating environments. The convergence of control scheme are analyzed, and the estimation accuracy of the observer for disturbances with different frequencies is demonstrated. Finally, an implementation to an automatic guided vehicle (AGV) is illustrated to verify the effectiveness of the proposed control scheme.
{"title":"Extended State Observer based Iterative Learning Control for Systems with Nonrepetitive Disturbances","authors":"Shiyan Li, Xuefang Li","doi":"10.1109/DDCLS58216.2023.10167248","DOIUrl":"https://doi.org/10.1109/DDCLS58216.2023.10167248","url":null,"abstract":"A novel extended state observer (ESO) based iterative learning control (ILC) scheme is investigated, including three channels, namely, feedforward, feedback, and disturbance rejection channels. The goal of this work is to achieve high-accuracy tracking of nonlinear systems in the presence of nonrepetitive disturbances under repetitive operating conditions. The ESO is used to estimate and offset the nonrepetitive disturbance in real time, which reduces the sensitivity of the controller to system parameters and operating environments. The convergence of control scheme are analyzed, and the estimation accuracy of the observer for disturbances with different frequencies is demonstrated. Finally, an implementation to an automatic guided vehicle (AGV) is illustrated to verify the effectiveness of the proposed control scheme.","PeriodicalId":415532,"journal":{"name":"2023 IEEE 12th Data Driven Control and Learning Systems Conference (DDCLS)","volume":"40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129987895","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-12DOI: 10.1109/DDCLS58216.2023.10166816
Qin Xiong, Huaiping Jin, Bin Wang, Haipeng Liu, Wangyang Yu
In batch processes, soft sensors frequently face the problem of irregular distributions between current and past data owing to variations in operating circumstances, as well as and poor model performing resulting from the absence with labels in the current data. This paper proposes a soft sensor method that is founded on dynamic multi-layer domain adaptation (DMDA). The method being proposed first training a convolutional neural network model with a substantial quantity of labeled data in the source domain, and subsequently use the obtained parameters as the beginning stage for the target model. Then, by utilizing multi-kernel maximum mean discrepancy (MK-MMD) and conditional embedding operator discrepancy (CEOD), the multi-layer convolutional neural network can effectively measure the difference in the overall (marginal) and specific (conditional) distributions between the source and target domains. Furthermore, the incorporation of an adaptive factor facilitates the dynamic adjustment of distribution weight, enabling precise fine-tuning of the target model. Finally, a regression model is established using the distribution-adapted historical data to achieve unsupervised soft sensor modeling. The substrate concentration in different fermentation tanks of the chlortetracycline fermentation process can be predicted through the utilization of the proposed approach. The experimental findings indicate that this method can accomplish tank-to-tank knowledge transfer, and significantly outperform traditional transfer learning-based soft sensor methods.
{"title":"A Soft Sensor Method based on Unsupervised Multi-layer Domain Adaptation for Batch Processes","authors":"Qin Xiong, Huaiping Jin, Bin Wang, Haipeng Liu, Wangyang Yu","doi":"10.1109/DDCLS58216.2023.10166816","DOIUrl":"https://doi.org/10.1109/DDCLS58216.2023.10166816","url":null,"abstract":"In batch processes, soft sensors frequently face the problem of irregular distributions between current and past data owing to variations in operating circumstances, as well as and poor model performing resulting from the absence with labels in the current data. This paper proposes a soft sensor method that is founded on dynamic multi-layer domain adaptation (DMDA). The method being proposed first training a convolutional neural network model with a substantial quantity of labeled data in the source domain, and subsequently use the obtained parameters as the beginning stage for the target model. Then, by utilizing multi-kernel maximum mean discrepancy (MK-MMD) and conditional embedding operator discrepancy (CEOD), the multi-layer convolutional neural network can effectively measure the difference in the overall (marginal) and specific (conditional) distributions between the source and target domains. Furthermore, the incorporation of an adaptive factor facilitates the dynamic adjustment of distribution weight, enabling precise fine-tuning of the target model. Finally, a regression model is established using the distribution-adapted historical data to achieve unsupervised soft sensor modeling. The substrate concentration in different fermentation tanks of the chlortetracycline fermentation process can be predicted through the utilization of the proposed approach. The experimental findings indicate that this method can accomplish tank-to-tank knowledge transfer, and significantly outperform traditional transfer learning-based soft sensor methods.","PeriodicalId":415532,"journal":{"name":"2023 IEEE 12th Data Driven Control and Learning Systems Conference (DDCLS)","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130241150","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-12DOI: 10.1109/DDCLS58216.2023.10166482
X. Xi, Pengqi Sun, C. Xing, Shengnan Li, Xincui Tian
An variational mode decomposition (VMD) has been applied in the field of harmonic detection, but the error will be large if the decomposition parameters are set artificially. To improve the accuracy of VMD in inter-harmonics detection, we need to determine the number of wolves, maximum number of iterations, convergence factor and other parameters, and then select component sample entropy function as the fitness function of the Grey Wolf algorithm. The variational mode decomposition can be utilized to extract the harmonic signal and choose a minimum envelope entropy weight as the best component. The Fourier transform is used to obtain the amplitude and frequency information of interharmonic signals. The simulation results show that the proposed method can effectively optimize the parameters and reduce the VMD decomposition error. Compared with empirical mode decomposition (EMD), complementary ensemble empirical mode decomposition (CEEMD) and empirical wavelet transform (EWT), the VMD with optimized parameters can significantly improve the accuracy of interharmonic detection and improve the accurate trace of accident source.
{"title":"A Parameter Optimized Variational Mode Decomposition Method for Harmonic and Inter-harmonic Detection","authors":"X. Xi, Pengqi Sun, C. Xing, Shengnan Li, Xincui Tian","doi":"10.1109/DDCLS58216.2023.10166482","DOIUrl":"https://doi.org/10.1109/DDCLS58216.2023.10166482","url":null,"abstract":"An variational mode decomposition (VMD) has been applied in the field of harmonic detection, but the error will be large if the decomposition parameters are set artificially. To improve the accuracy of VMD in inter-harmonics detection, we need to determine the number of wolves, maximum number of iterations, convergence factor and other parameters, and then select component sample entropy function as the fitness function of the Grey Wolf algorithm. The variational mode decomposition can be utilized to extract the harmonic signal and choose a minimum envelope entropy weight as the best component. The Fourier transform is used to obtain the amplitude and frequency information of interharmonic signals. The simulation results show that the proposed method can effectively optimize the parameters and reduce the VMD decomposition error. Compared with empirical mode decomposition (EMD), complementary ensemble empirical mode decomposition (CEEMD) and empirical wavelet transform (EWT), the VMD with optimized parameters can significantly improve the accuracy of interharmonic detection and improve the accurate trace of accident source.","PeriodicalId":415532,"journal":{"name":"2023 IEEE 12th Data Driven Control and Learning Systems Conference (DDCLS)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129227813","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Due to the scarcity of fault samples and the weakness of processing higher-order interactive information, the most existing intelligence methods fail to achieve the optimal effect in fault diagnosis. To address these problems, a time-frequency hypergraph neural network-based fault diagnosis method is proposed. In the proposed network, the limited data is initially segmented using the sliding window mechanism to obtain a set of time-domain signal instances. Additionally, the Fast Fourier Transform (FFT) is applied to each signal instance to extract corresponding frequency-domain signals, so as to capture more fault-sensitive features. Subsequently, a two-layer convolutional neural network is used to extract fault-attention features from both the time and frequency domain signals. Also, in order to reduce computational complexity, the time-frequency domain features are adaptively stacked based on a self-attention mechanism. Furthermore, a feature similarity graph is constructed for the time-frequency domain features using a k-nearest neighbor algorithm. This graph is then input into the hypergraph neural network (HGNN) to obtain the final diagnosis results. One comparative experiment shows that the proposed method not only mitigates the performance degradation caused by limited samples and noisy environments, but also effectively leverages the higher-order interaction information among nodes in the hypergraph.
{"title":"Time-frequency Hypergraph Neural Network for Rotating Machinery Fault Diagnosis with Limited Data","authors":"Haobin Ke, Zhi-wen Chen, Jiamin Xu, Xinyu Fan, Chao Yang, Tao Peng","doi":"10.1109/DDCLS58216.2023.10167156","DOIUrl":"https://doi.org/10.1109/DDCLS58216.2023.10167156","url":null,"abstract":"Due to the scarcity of fault samples and the weakness of processing higher-order interactive information, the most existing intelligence methods fail to achieve the optimal effect in fault diagnosis. To address these problems, a time-frequency hypergraph neural network-based fault diagnosis method is proposed. In the proposed network, the limited data is initially segmented using the sliding window mechanism to obtain a set of time-domain signal instances. Additionally, the Fast Fourier Transform (FFT) is applied to each signal instance to extract corresponding frequency-domain signals, so as to capture more fault-sensitive features. Subsequently, a two-layer convolutional neural network is used to extract fault-attention features from both the time and frequency domain signals. Also, in order to reduce computational complexity, the time-frequency domain features are adaptively stacked based on a self-attention mechanism. Furthermore, a feature similarity graph is constructed for the time-frequency domain features using a k-nearest neighbor algorithm. This graph is then input into the hypergraph neural network (HGNN) to obtain the final diagnosis results. One comparative experiment shows that the proposed method not only mitigates the performance degradation caused by limited samples and noisy environments, but also effectively leverages the higher-order interaction information among nodes in the hypergraph.","PeriodicalId":415532,"journal":{"name":"2023 IEEE 12th Data Driven Control and Learning Systems Conference (DDCLS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129418186","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-12DOI: 10.1109/DDCLS58216.2023.10166696
Wancheng Yue, Junsheng Ren, Weiwei Bai
This paper proposed a method of online non-parameter identification of nonlinear ship motion systems. Firstly, we use Mariner to generate a certain amount of ship motion data to train the LWPR model. Then the ship travels along a set track. During this process, the sensors continuously obtain the distance, radial velocity and azimuth of the ship relative to the ship, and then completes the construction of simulation data. Next, the performance of the algorithm is verified which uses the Kalman filtering framework. Finally, the estimated value is further used for updating the LWPR model to achieve the purpose of online learning, and the updated model will be used for the next prediction. The experimental results show that the online modeling and tracking method proposed in this paper has higher tracking accuracy than the parameter estimation techniques.
{"title":"Online non-parametric modeling for ship maneuvering motion using local weighted projection regression and extended Kalman filter","authors":"Wancheng Yue, Junsheng Ren, Weiwei Bai","doi":"10.1109/DDCLS58216.2023.10166696","DOIUrl":"https://doi.org/10.1109/DDCLS58216.2023.10166696","url":null,"abstract":"This paper proposed a method of online non-parameter identification of nonlinear ship motion systems. Firstly, we use Mariner to generate a certain amount of ship motion data to train the LWPR model. Then the ship travels along a set track. During this process, the sensors continuously obtain the distance, radial velocity and azimuth of the ship relative to the ship, and then completes the construction of simulation data. Next, the performance of the algorithm is verified which uses the Kalman filtering framework. Finally, the estimated value is further used for updating the LWPR model to achieve the purpose of online learning, and the updated model will be used for the next prediction. The experimental results show that the online modeling and tracking method proposed in this paper has higher tracking accuracy than the parameter estimation techniques.","PeriodicalId":415532,"journal":{"name":"2023 IEEE 12th Data Driven Control and Learning Systems Conference (DDCLS)","volume":"104 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128505365","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：