Transactions of China Electrotechnical Society  2019, Vol. 34 Issue (16): 3311-3321    DOI: 10.19595/j.cnki.1000-6753.tces.180962
Current Issue| Next Issue| Archive| Adv Search |
Insulator State Detection of Convolutional Neural Networks Based on Feedback Mechanism
Zhang Qian1, Wang Jianping1, Li Weitao1,2
1. School of Electric Engineering and Automation Hefei University of Technology Hefei 230009 China;
2. State Key Laboratory of Synthetical Automation for Process Industries Northeastern University Shenyang 110004 China

Download: PDF (41686 KB)   HTML (1 KB) 
Export: BibTeX | EndNote (RIS)      
Abstract  As an important equipment on the transmission line, the conditions of insulators directly affect the safe operation of power grid. Therefore, it is important and necessary to detect the insulator state automatically and accurately. However, existing detection models with constant feature space have drawbacks for distinctive samples, and it is complex to extract insulator image features. Thus, a method for insulator state detection based on convolutional neural network and feedback mechanism is explored in this paper, which imitates the human cognition process with repetitive comparison and inference from simple to fine. Firstly, according to the characteristics of the insulator samples, the network structure of LeNet-5 is improved by employing stochastic configuration networks (SCNs) and the feedback mechanism, and the alternate optimization strategy is adopted to optimize the parameters of the convolution neural network. Secondly, based on the entropy theory, the semantic error entropy measurement index is established to evaluate the uncertainty of the insulator detection results in real time to overcome the posterior evaluation. Finally, according to the evaluation results of semantic error information entropy measurement index, the feedback mechanism is constructed to adjust the size and number of the convolution kernels to extract the refined features for the repeated comparison and inference. The experimental results show that the improved convolution neural network model has a superior accuracy in insulator state detection.
Key wordsInsulator state detection      convolutional neural network      feedback mechanism      semantic error information entropy     
Received: 01 June 2018      Published: 02 September 2019
PACS: TM216  
Service
E-mail this article
Add to my bookshelf
Add to citation manager
E-mail Alert
RSS
Articles by authors
Zhang Qian
Wang Jianping
Li Weitao
Cite this article:   
Zhang Qian,Wang Jianping,Li Weitao. Insulator State Detection of Convolutional Neural Networks Based on Feedback Mechanism[J]. Transactions of China Electrotechnical Society, 2019, 34(16): 3311-3321.
URL:  
https://dgjsxb.ces-transaction.com/EN/10.19595/j.cnki.1000-6753.tces.180962     OR     https://dgjsxb.ces-transaction.com/EN/Y2019/V34/I16/3311
Copyright © Transactions of China Electrotechnical Society
Supported by: Beijing Magtech