Research And Design Of PID Control System Based On BP Neural Network

PID controller is a simple,effective and reliable control algorithm widely used in industry.However,the control effects of traditional PID controllers are not good when dealing with the complex nonlinear controlled systems.In order to improve the defects of traditional PID to meet higher industrial requirements,it is necessary to design an improved PID controller with better performance.The artificial neural network technology not only has the self-learning ability,but also has nonlinear mapping ability.How to make more effective use of the advantages of neural network and traditional PID,to achieve better results in the field of industrial control,has become a research hotspot.Based on the above,the specific content of this article is as follows:Firstly,the paper analyzes the advantages and disadvantages of traditional PID controllers,deduces the formula of digital PID algorithm,and introduces several improved PID controllers and their theoretical basis.Secondly,this paper introduces the single neuron PID controller,explains the selection of the activation function,and deduces the learning rules such as Hebb and Delta.Furthermore,the paper introduces the concept of BP neural network,and gives the algorithm derivation.Finally,and most importantly,based on the above,this paper takes the motor simplified model as an example for the complex nonlinear controlled object that is difficult to control with traditional PID.In order to improve its inherent accuracy,such as poor positioning accuracy and poor initial stability,the neural network is combined with traditional PID and improved and optimized measures are proposed.In this paper,a modified BP neural network PID controller with higher performance is designed to achieve the goal of improving motor performance.In the MATLAB environment,the designed PID is simulated and debugged.The simulation results show that compared with the single neuron PID,the improved BP neural network PID controller reduces the system's rise time by 0.005 s,the overshoot by 1.5%,the adjustment time by 0.088 s,and the error peak range by ±0.029.These simulation results show that the designed PID control performance has been greatly improved.And the result can be further extended to more complex controlled systems.After verifying the correctness and validity of the algorithm,this paper further designs and discusses the implementation of the proposed PID on FPGA.