Research On Linear Motor Control System Based On Deep Reinforcement Learning Algorithm

Precision air suspension platform is often used as the carrier platform of precise movement in the field of precision machining.In order to achieve high-precision operation of precision platform,permanent magnet synchronous linear motor(PMSLM),as the actuator of precision air suspension platform,has the dynamic performance requirements of low overshoot and fast adjustment for its control system.Depth deterministic strategy gradient(DDPG)algorithm is a data-driven deep reinforcement learning algorithm.It can directly design and analyze the controller according to the input and output data of the system,and realize the optimal control of the system according to the given reward.This kind of deep reinforcement learning algorithm is introduced into the motor control system to meet the dynamic performance requirements of linear motor with its excellent self-learning and self-adaptive ability.This thesis takes the precision air suspension platform as the research background,and takes the control system of the permanent magnet synchronous linear motor as the research object,and deeply studies a linear motor control system based on deep reinforcement learning algorithm.Firstly,the control system of the PMSLM is analyzed,and the mathematical model of the linear motor is established according to the actual structure of the permanent magnet synchronous linear motor and the principle of vector control;then,the fuzzy PID control system model of the linear motor is built on the Simulink platform,and the simulation is carried out.Secondly,in the case of linear motors running at high speed,in order to improve the anti-jamming capability of the control system and achieve low overshoot dynamic performance requirements,an adaptive controller based on the DDPG algorithm is proposed.The adaptive controller is compared with the fuzzy PID controller on Simulink platform.The simulation results show that the DDPG adaptive controller has the advantages of short adjustment time,strong adaptive ability and good following effect,etc.,which improves the dynamic performance and anti-interference ability of the linear motor control system.Finally,due to the limitation of the existing precision air suspension hardware platform,the precision air suspension platform is improved to verify the effectiveness of the deep reinforcement learning algorithm on the hardware platform.The DDPG adaptive controller is run in the upper computer,and then the control signal is converted and transmitted through the STM32F103 development board.Through the actual measurement operation on the hardware platform,it is verified that the control system has better adaptability and excellent dynamic response performance.