Simulation Research Of PID Neural Network Controller Based On Artificia Fish Swarm Algorithm

With the research of intelligent control, the application of artificial intelligent techniques is used more and more in the control field.The PID neural network(PIDNN) is a new type of neural network with the metrits of PID control and neural network.It’s esay to design the PID neural network controllers for all kinds of systems. But the BP training algorithm restrains the PID neural network’s wide applications to control field,because the PIDNN controller can’t get good wights.In veiw of the problem,in this paper artificia fish swarm algorithm(AFSA) is used to trainning the PIDNN controllers, and did the control simulation for different kinds of systems. the main research work of this dissertation as follows:(l)We analyzed the advantages and disadvantages of the artificia fish swarm algorithm based on the in-depth study of the algorithm. For the defects of the algorithm,the paper addresses some improvement strategies with the summary the other scholar’s ideas. The simulation test verify the effectiveness of the strategies.(2)We introduced the PID neural network, analyzed the structure and algorithm of the PIDNN controller, and then we proposal some improvement strategies.We designed the PIDNN controllers for the different kinds of systems.(3)For the defects of the PIDNN controller with the BP training algorithm, we proposal the idea of using the artificia fish swarm algorithm to training the weights of PIDNN controller, and the process of training is summarized.(4) The simulation of PIDNN controllers based on AFSA for the single-variable system, square process and non-square process are presented in the MATLAB environmnet.The simulation of PID controllers based on AFSA, PIDNN controllers based on BP algorithm, indicate that the AFSA resolve the weights of the PIDNN controller, makes it better in perormances. The simulation verify the validity of the method. The AFSA extend the application range of the PIDNN controllers.