Junyou Shi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Junyou Shi has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Yi Deng
School of Reliability and Systems Engineering, Beihang University, Beijing, 100191, China
2 shared publications - Zili Wang
State Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang University, Hangzhou, China
2 shared publications - Meng Shi
School of Reliability and System Engineering, Beihang University, Xueyuan Road 37, 100191 Beijing, China
1 shared publication Qingjie He
1 shared publication- Qingjie He
School of Reliability and Systems Engineering, Beihang University, 37, Xueyuan Road, Beijing, China
1 shared publication - Xie-Gui Lin
School of Reliability and System Engineering, Beihang University, Xueyuan Road 37, 100191 Beijing, China
1 shared publication
Latest Publications
- Open Access
A Diagnosis Method for Noise and Intermittent Faults in Analog Circuits Based on the Fusion of Multiscale Fuzzy Entropy Features and Amplitude Features
Intermittent faults occur randomly, last for short durations, and ultimately lead to permanent failures, threatening the safety and stability of analog circuits. Additionally, these faults are often hard to differentiate from noise-induced anomalies, resulting in incorrect disassembly and complicating circuit maintenance. To address these challenges, we propose a novel fault diagnosis method. The method uses an adjustable sliding window to extract multiscale fuzzy entropy features, mitigating the impact of normal data on entropy calculations for intermittent faults. The coarse granulation strategy of sliding point by point is applied to avoid information loss in short time series. The raw signal is then segmented and transformed into four statistical features, which are fused into comprehensive amplitude features via a self-attention mechanism. This comprehensive feature better captures amplitude variations than individual statistical features. Finally, the two features are fed into a convolutional neural network for diagnosis. The method is applied to two typical analog circuits. Ablation studies confirmed its effectiveness. Although the proposed method does not have the lowest diagnostic cost and the fastest detection time, the differences with state-of-the-art methods are minimal, and the proposed method achieves higher classification accuracy. Taken together, these findings demonstrate the superiority of the proposed method.
Open AccessA Diagnosis Method for Noise and Intermittent Faults in Analog Circuits Based on the Fusion of Multiscale Fuzzy Entropy Features and Amplitude Features
Intermittent faults occur randomly, last for short durations, and ultimately lead to permanent failures, threatening the safety and stability of analog circuits. Additionally, these faults are often hard to differentiate from noise-induced anomalies, resulting in incorrect disassembly and complicating circuit maintenance. To address these challenges, we propose a novel fault diagnosis method. The method uses an adjustable sliding window to extract multiscale fuzzy entropy features, mitigating the impact of normal data on entropy calculations for intermittent faults. The coarse granulation strategy of sliding point by point is applied to avoid information loss in short time series. The raw signal is then segmented and transformed into four statistical features, which are fused into comprehensive amplitude features via a self-attention mechanism. This comprehensive feature better captures amplitude variations than individual statistical features. Finally, the two features are fed into a convolutional neural network for diagnosis. The method is applied to two typical analog circuits. Ablation studies confirmed its effectiveness. Although the proposed method does not have the lowest diagnostic cost and the fastest detection time, the differences with state-of-the-art methods are minimal, and the proposed method achieves higher classification accuracy. Taken together, these findings demonstrate the superiority of the proposed method.
- 7 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Components, Packaging and Manufacturing Technology
- Vol. 11 (1) , 71-82
A Transfer Learning LSTM Network-Based Severity Evaluation for Intermittent Faults of an Electrical Connector
Traditional machine learning methods work well under a general assumption that the training data and the testing data should be drawn from the same distribution. However, this assumption does not hold in many real world applications. For two structure-similar but specification-different connectors with intermittent fault, one connector has sufficient intermittent fault data with severity labeled but another connector has no labeled intermittent fault data. In this situation, the severity evaluation model trained by the samples from the former connector will have large evaluation errors for the samples from the latter connector. For realizing the transfer of the intermittent fault severity evaluation model between two similar but different connectors, this paper proposes a transfer learning Long Short Term Memory (LSTM) network based severity evaluation method for intermittent faults of an electrical connector. Conventional transfer learning methods with Maximum Mean Discrepancy (MMD) have two shortcomings: 1) MMD only considers the mean discrepancy between the source and target domains; 2) MMD only measures the marginal distribution discrepancy of the source and target domains but ignored the conditional distribution discrepancy of different labels. Therefore this paper proposes a modified MMD metrics which not only considers the higher order statistics but also the label discrepancy information to enhance the severity evaluation accuracy. Finally, this transfer learning model is applied in the intermittent fault experiment of three electrical connectors, and the effectiveness of the proposed method is verified using comparisons with four state-of-art transfer learning methods.
- 7 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Components, Packaging and Manufacturing Technology
- Vol. 11 (1) , 71-82
A Transfer Learning LSTM Network-Based Severity Evaluation for Intermittent Faults of an Electrical Connector
Traditional machine learning methods work well under a general assumption that the training data and the testing data should be drawn from the same distribution. However, this assumption does not hold in many real world applications. For two structure-similar but specification-different connectors with intermittent fault, one connector has sufficient intermittent fault data with severity labeled but another connector has no labeled intermittent fault data. In this situation, the severity evaluation model trained by the samples from the former connector will have large evaluation errors for the samples from the latter connector. For realizing the transfer of the intermittent fault severity evaluation model between two similar but different connectors, this paper proposes a transfer learning Long Short Term Memory (LSTM) network based severity evaluation method for intermittent faults of an electrical connector. Conventional transfer learning methods with Maximum Mean Discrepancy (MMD) have two shortcomings: 1) MMD only considers the mean discrepancy between the source and target domains; 2) MMD only measures the marginal distribution discrepancy of the source and target domains but ignored the conditional distribution discrepancy of different labels. Therefore this paper proposes a modified MMD metrics which not only considers the higher order statistics but also the label discrepancy information to enhance the severity evaluation accuracy. Finally, this transfer learning model is applied in the intermittent fault experiment of three electrical connectors, and the effectiveness of the proposed method is verified using comparisons with four state-of-art transfer learning methods.
- 1 December 2020
- Published by Elsevier in Applied Soft Computing
- Vol. 97, 106835
An adaptive new state recognition method based on density peak clustering and voting probabilistic neural network
- 1 December 2020
- Published by Elsevier in Applied Soft Computing
- Vol. 97, 106835
An adaptive new state recognition method based on density peak clustering and voting probabilistic neural network
- Open Access
- 1 October 2012
- Published by Polish Academy of Sciences Chancellery in Metrology and Measurement Systems
- Vol. 19 (3) , 509-520
A key metric and its calculation models for a continuous diagnosis capability base dependency matrix
This paper is devoted to measuring the continuous diagnosis capability of a system. A key metric and its calculation models are proposed enabling us to measure the continuous diagnosis capability of a system directly without establishing and searching the sequential fault tree (SFT) of the system. At first a description of a D matrix is given and its metric is defined to determine the weakness of a continuous diagnosis. Then based on the definition of a sequential fault combination, a sequential fault tree (SFT) is defined with its establishment process summarized. A key SFT metric is established to measure the continuous diagnosis capability of a system. Two basic types of dependency graphical models (DGMs) and one combination type of DGM are selected for characteristics analysis and establishment of metric calculation models. Finally, both the SFT searching method and direct calculation method are applied to two designs of one type of an auxiliary navigation equipment, which shows the high efficiency of the direct calculation method.
- 1 October 2012
- Published by Polish Academy of Sciences Chancellery in Metrology and Measurement Systems
- Vol. 19 (3) , 509-520
Open AccessA key metric and its calculation models for a continuous diagnosis capability base dependency matrix
This paper is devoted to measuring the continuous diagnosis capability of a system. A key metric and its calculation models are proposed enabling us to measure the continuous diagnosis capability of a system directly without establishing and searching the sequential fault tree (SFT) of the system. At first a description of a D matrix is given and its metric is defined to determine the weakness of a continuous diagnosis. Then based on the definition of a sequential fault combination, a sequential fault tree (SFT) is defined with its establishment process summarized. A key SFT metric is established to measure the continuous diagnosis capability of a system. Two basic types of dependency graphical models (DGMs) and one combination type of DGM are selected for characteristics analysis and establishment of metric calculation models. Finally, both the SFT searching method and direct calculation method are applied to two designs of one type of an auxiliary navigation equipment, which shows the high efficiency of the direct calculation method.