The Study Of PMSM Controller Based On Neural Network Sliding Mode Control

Three steps,breaking through ten fields that is issued in "made in China 2025" point out:"Focus on improving reliability and accuracy retention,development of high-end digital control system,servo motor and other key functional components and key application software to accelerate the realization of industrialization." According to the national development plan,it is urgent to develop a more advanced servo motor control system,many experts and scholars apply various advanced control strategies to permanent magnet synchronous motor servo system to meet the high quality requirements of the system.Sliding mode control has strong robustness,fast response,but poor learning ability,and the system squats on buffeting problems can not be eliminated.Neural network can approximate any nonlinear system,and has excellent learning ability,but its response speed is very slow.In view of the characteristics of both,the paper combines these two advanced control strategies and applies them to speed loop control of permanent magnet synchronous motor to improve the performance of servo system.Firstly,the decoupling equation of PMSM is deduced from the principle of coordinate transformation.Then,use the Vector control strategy that is id=O.Finally combination of space voltage vector control technology to set up the double closed loop control system of PMSM.Subsequently,the sliding mode variable structure control is introduced from two aspects of algorithm principle and quality characteristic,and derived the algorithm of conventional sliding mode control for speed loop.In this article,a neural network is used to design the system identifier for on-line identification of the parameters of the controlled object,and another neural network is combined with the sliding mode variable structure controller.Thus the response speed,convergence and stability of the system are improved.Finally,the article combines the advantages of neural network and sliding mode control,then proposes an ideal sliding mode surface.Also the neural network is used to keep the system in full sliding mode.Finally,the buffeting of the system is diminished by adjusting the parameters of the controller online with neural network,and the robustness of the motor servo system is enhanced.