Research On Position Servo Control Of Voice Coil Motor Based On Deep Learning Of Optimal Data

Deep learning has attracted more and more attention as a popular research direction in recent years.In the field of control,traditional control methods need to be based on the model of the controlled object,but they do not have a good grasp of the laws.Deep learning is a method based on data learning.By learning the rules of data,it can achieve good control performance.This article mainly discusses the position servo control of a rotary voice coil motor.A position servo control system based on deep neural network was designed to enable precise position control under different position requirements.First of all,it is necessary to obtain the data for training the neural network and start with the voice coil motor of the controlled object.The structure and mathematical model of the rotary voice coil motor are introduced.Based on the actual position control target,the optimized object is analyzed and the optimized model of the rotary voice coil motor is established.The CPLEX optimization solver obtains the optimized control data of the rotary voice coil motor under different working conditions.The characteristics of the optimized data are analyzed combined with the working conditions and parameters of rotary voice coil motor.Second,Compare the control performance of the three neural networks and select the optimal network as the focus of research.The network structure and main algorithm characteristics of the three neural networks of BP neural network,Elman neural network and RBF neural network are introduced.Three kinds of neural networks were designed for selecting network layers,the number of neurons,the transfer function,the training function,and learning function.Based on the characteristics of the structure and the algorithm,different designs among the three neural networks are compared,and the three neural networks are learned and trained using the designed network structure.The simulation model of the position servo control system of voice coil motor based on neural network was built,and the control effects of three neural networks were compared and analyzed.Then,it mainly analyzes the system controlled by BP neural network.The characteristics of the distinction and control between shallow network and deep network are discussed.The influence of the number of hidden neurons in the neural network on the control effect is compared.The selection of the input variables in the neural network and the influence of the parameter changes on the position control effect are analyzed.The changes brought by speed filtering to position control are researched,and the optimal network was selected for further analysis.Finally,an experimental platform was set up based on the position servo control system of the rotating voice coil motor.The relevant experimental research and experimental data analysis were performed.It was proved that the data-based control method only learns limited data,but it can learn the rules in the data and achieve a good control effect.