Study On Algorithm Improvement Of BP Neural Networks And Its Application In PID Control

Artificial neural network (ANN), as an important part of artificial intelligence, has great potential in application. After introducing the development, status quo, basic theory of neural network and its application to automatic control, this thesis mainly studies the structures and algorithms of BP neural network and its application to PID control.BP neural network is the most used neural network at present. It has unique approximation ability and simple structure, and it is a neural network with good performance. So it has particular advantages when applied in PID control.This paper studies the theory of BP neural network, analyzes the advantages and disadvantages of several popular training algorithms. To deal with the defects of the steepest descent in slowly converging and easily immerging in partial minimum frequently, after analyzing the linear hunting method developed by Fletcher and Reeves, the improved conjugate gradient algorithm is brought forward to solve the problem. This paper analyzes the algorithm deeply in theory, introduces the idea and process. Then the BP neural network trained by this algorithm is applied into function approximation. The results show that this algorithm improves the convergence of training process and achieves excellent identification effect.Applying the BP neural network in PID control can efficiently overcome the limitations of badness of parameter adjusting and poor performance when the plant has nonlinearity, time-varying uncertainty and difficulty in setting up the accurate model. This paper studies the structure and algorithm of PID controller based on BP neural network, applies the improved conjugate gradient to neural network PID controller, and proposes a new type of PID control method based on BP neural network. The simulation results show that this improved algorithm not only increases the convergence speed in the training process, but also adjusts the PID controller parameters on line, which has rather strong capabilities of adaptive and self-study. So it has better performance.