Title: Research on operation stability evaluation of industrial automation system based on improved deep learning
Authors: Bo Peng
Addresses: Department of Development, Shenzhen Haiweite Energy-Saving Science and Technology Co., Ltd., Shenzhen, 518172, China
Abstract: In order to overcome the problems of low evaluation accuracy, long evaluation time and high data extraction error of traditional methods, an evaluation method of industrial automation system operation stability based on improved deep learning is proposed. This paper analyses the key indicators of industrial automation system operation stability evaluation, activates the sample data with the help of binary cross entropy function, and obtains the partial derivative of artificial neural network to complete the improvement of artificial neural network. The running characteristics of industrial automation system are extracted, and the feature data are de-noising with the help of self-encoder. These data are input into the improved artificial neural network, and the evaluation results are output. The experimental results show that the highest evaluation accuracy of the proposed method is about 96%, the evaluation time is less than 0.6 s, and the error of feature data extraction is only 2.1%.
Keywords: improved deep learning; industrial automation system; stability evaluation; binary cross entropy function; self-encoder.
International Journal of Manufacturing Technology and Management, 2022 Vol.36 No.2/3/4, pp.141 - 153
Received: 22 Dec 2020
Accepted: 03 Jun 2021
Published online: 30 Jun 2022 *