PID Controller Based Adaptive GA and Neural Networks

A self-tuning PID controller based on adaptive genetic algorithm (AGA) and neural networks is presented. AGA optimizes not only the initial weights of the BP neural networks (BPNN) which optimizes parameters of PID, but also the optimum values of the following radial basis function neural networks (RBFNN) parameters: centers, variance and weights of the output layer. RBFNN identifies the Jacobian information of the controlled plant. The influence on the control performance is solved which results from the initial parameters of BPNN and RBFNN. The result of the simulation shows that the method can improve the robust performance of the control system

[1]  Sigeru Omatu,et al.  Self-tuning PID control by neural-networks , 1993, Proceedings of 1993 International Conference on Neural Networks (IJCNN-93-Nagoya, Japan).

[2]  Xianyang Jiang,et al.  Controlling chaos by RBF neural network based on GA optimization , 2004, Fifth World Congress on Intelligent Control and Automation (IEEE Cat. No.04EX788).

[3]  Liu Sheng.  An improved adaptive genetic algorithm and its application in function optimization , 2007 .

[4]  Sigeru Omatu,et al.  PID Control by Neural Networks , 1993 .

[5]  Lalit M. Patnaik,et al.  Adaptive probabilities of crossover and mutation in genetic algorithms , 1994, IEEE Trans. Syst. Man Cybern..

[6]  Chen Guo-ding.  PID Controller with Neural Nets Optimizing Parameters and Its Stimulation , 2005 .