Research On Parameter Optimization Of PID Control Algorithm Based On BP Neural Network

In today's society,traditional strategy control methods have become very mature.Among them,PID(Proportional Integral Derivative)control algorithm is most commonly used in the field of industrial process control.With the vigorous development of control theory,people have gradually combined cybernetics with other disciplines to form some new control theories.Neural networks are typical examples of them.Because neural networks have strong self-learning and approximation capabilities for non-linear functions,neural networks can be perfectly applied to PID control algorithms.Taking this as the starting point,this paper introduces the neural network neuron model and its network structure,derives its algorithm,and analyzes the advantages and disadvantages of BP neural network.Then introduced the traditional PID controller.On this basis,a BP neural network PID controller is designed,and the network structure,activation function,initial weight and learning rate of the BP neural network PID are selected one by one,and the BP neural network PID algorithm is derived.In order to solve the problems of slow convergence,oversaturation and network fluctuation and oscillation of PID algorithm of BP neural network,the method of improving learning rate and adding momentum correction factor was used to solve it.The MATLAB simulation tool was used for targeted simulation.The results of the simulation It shows that after the method of improving learning rate and adding momentum factor,the system performance has been effectively improved.Then,taking the double closed-loop DC speed control system as the column,the optimized BP neural network PID controller is used in industrial control.MATLAB / simulink simulation software is used to build the double closed-loop DC speed control system model,and BPPID speed adjustment is added to it.The simulation compares the speed control system with optimized BPPID and the speed control system without optimized BPPID.The results show that the system with BP neural network speed PID regulator has small overshoot,fast dynamic response,strong anti-interference ability,and adjustment speed.Quickly,the control effect has been effectively improved.