Research On RBF-PID Control Based On Visual Cricket System

Ball & plate system is a typical multivariable,uncertain and strongly coupled nonlinear system.It has been used by many scholars at home and abroad to study the experimental platform of nonlinear control system.In this paper,we mainly study the position control and trajectory tracking control of Ball & plate system based on adaptive control algorithm of RBF neural network expect to improve the accuracy of position control and trajectory tracking.Firstly,this paper introduces the hardware structure and working principle of GBP2001 cricket system in detail,and then analyzes the dynamics of the ball in the ball plate.Secondly,Lagrange equation are used to establish a complete mathematical model of the Ball & plate system.Lastly,this paper presents two control models of angular acceleration control and Angle control,and gives the reason of using Angle control model,which proves that the model is controllable in a certain range.Secondly,the paper introduces the RBF neural network algorithm,the theoretical basis of PID controller and the principle of tuning PID parameters of RBF neural network.For PID control(original system),the control time is long and the control accuracy is not high,this article first by using RBF neural network structure is simple,the advantages of available approximate any nonlinear function combined with PID control algorithm,realization of PID parameters self-tuning control,and give full play to the RBF neural network nonlinear mapping and self-learning ability two advantages.The self-adaption PID control of the Ball & plate system is realized.Then,for the problem of the RBF neural network of "near excitation remote inhibition",based on RBF-PID parameter adjustment adopts gradient descent method,introducing the acceleration rate to accelerate the system response speed and shorten the control time.Finally,the momentum factor was added in the parameter adjustment of RBF neural network,to reduce system oscillation;Using the LM(Levenberg Marquardt)algorithm instead of gradi ent descent method for Online optimization of PID parameters,to speed up the system response.The simulation experiment of RBF-PID control algorithm and its improved algorithm has verified the feasibility of applying the control algorithm to the Ball & plate system.Lastly,the improved algorithm of this paper is applied to the GBP2001 Ball & plate system to complete the position control and trajectory tracking control of the Ball & plate system.The experimental results show that the improved algorithm is feasible and robust,not only can achieve the control requirement,but also the control precision is high.The improved algorithm can reduce the influence of uncertainty and disturbance on the control precision of Ball & plate system.The control effect of system is very excellent.