文献类型：期刊文献

英文题名：Model Free Adaptive Control Algorithm based on GRU network

作者：Sun, Jinggao[1]; Chen, Xianfeng[1]; Su, Guanghao[1]; Pan, Hongguang[2]

机构：[1] Key Laboratory of Advanced Control and Optimization for Chemical Process, Ministry of Education, East China University of Science and Technology, Shanghai, 200237, China; [2] School of Electrical and Control Engineering, Xi'An University of Science and Technology, Xi'an, 710049, China

年份：2021

起止页码：4758

外文期刊名：Proceedings of the 33rd Chinese Control and Decision Conference, CCDC 2021

收录：EI(收录号：20220911715907)

基金：National Nature Science Foundation under Grant No.62003139.

语种：英文

外文关键词：Learning systems - Errors - Adaptive control systems - Backpropagation - Multivariable systems

摘要：The application of traditional adaptive control algorithm usually depends on the precise mathematical model of process, but it is difficult to establish a mathematical model for a dynamic process. The existing Model Free Adaptive (MFA) control algorithm structure is based on Back Propagation(BP) Neural Network, which does not fully take into account the continuous characteristic relationship between the input timing error sequences. According to this actual problem, an improved MFA control algorithm based on Gated Recurrent Unit (GRU) network is proposed in this paper, compared with BP network, GRU network has more advantages in processing timing series. It fully considers the information of error sequence, and more appropriate manipulated variable could be generated automatically to satisfy the demand of an open-loop stable and controllable single-variable, multivariable or large delay process system without the need for complicated manual adjustments, quantitative knowledge of the process or the identifier of the controlled system and learning process. Simulation results demonstrate that the GRU-MFA learning algorithm performs better in terms of stability, response speed and adaptability than the BP-MFA algorithm without human intervention. ? 2021 IEEE.