APPLICATION OF SELF-TUNING PID CONTROL BASED ON DIAGONAL RECURRENT NEURAL NETWORK IN CRYSTALLIZATION PROCESS
In crystallization process there are strong coupling between the temperature control and the level control in the tank, and the parameter is time-varying, so it is difficult to apply general PID control. The self-adjusting PID control method based on diagonal recurrent neural network (DRNN) is introduced in this paper. According to the influence of objects parameter to system output performance, the DRNN can auto-adjust its weights to vary kP, kI and kD. The Simulation results show that the presented control system has quick response speed and strong adaptive capability.
Crystallization process DRNN Decoupling Control Simulation
ZHE-YING SONG CHAO-YING LIU XUE-LING SONG
College of Electrical Engineering & Informational Science, Hebei University of Science and Technology, Shijiazhuang 050054, Hebei, China
国际会议
2006 International Conference on Machine Learning and Cybernetics(IEEE第五届机器学习与控制论坛)
大连
英文
365-369
2006-08-13（万方平台首次上网日期，不代表论文的发表时间）