Optimal Design Of Differential Evolution Algorithm For Inverted Pendulum Fuzzy Neural Network Controller

Inverted pendulum system is a typical controlled model.It has the characteristics of under-redundancy,high-order,multi-state,and instability.It is an important tool for verifying the effectiveness of control methods.The study of its control method has profound theoretical and practical significance.Since the birth of inverted pendulum system,the study of its control method has almost included most of the existing control theory methods.With the rapid development of artificial intelligence technology,intelligent control theory represented by neural networks and evolutionary algorithms has gradually become an important research direction in the field of control.This paper takes the three-stage inverted pendulum system as the research object,analyzes its own structural characteristics and composition,and makes a qualitative analysis of the characteristics of the system through the mathematical model.For the problems of the error in the mathematical modeling of the three-stage inverted pendulum system and the design of adaptive controller,the fuzzy neural network and differential evolution algorithm are used to design the identifier and controller.The specific work content is as follows:(1)A three-stage inverted pendulum system identifier is designed using a fuzzy neural network.The identifier is constructed using a combination of multiple fuzzy neural networks to avoid dimensional explosion problems.The function model of the neural network identifier was constructed in MATLAB.Using the input and output data samples of a three-stage inverted pendulum system to train the fuzzy neural network identifier,a fuzzy neural network identification model of the system was obtained.(2)Aiming at the errors in mathematical modeling and the problem of the fuzzy neural network falling into a local optimal problem,this paper proposes a strategy of using differential evolution algorithm to optimize the fuzzy neural network controller to adaptive control the inverted pendulum system.Different three-stage inverted pendulum state space equations are used to the simulation of controller optimize.The experimental results show that this control strategy can achieve stable control of the three-stage inverted pendulum system under the condition that the initial system modeling error exists.(3)This paper starts with three aspects to improve the differential evolution algorithm.A dynamic classification differential evolution algorithm is proposed.This algorithm adopts the dynamic multi-group classification variation method,which makes the algorithm get a good balance between global search and local optimization.Finally,the fuzzy neural network controller is optimized by using this algorithm.The results show that the algorithm has significantly improved the search ability and convergence speed compared with the traditional differential evolution algorithm.The fuzzy neural network controller achieves ideal results.