永磁同步电机BP神经网络 智能PID滑模观测矢量控制算法  

BP Neural Network-intelligent PID Synovial Observation Vector Control Algorithm for Permanent Magnet Synchronous Motor

作　　者：郑瑞 张继祥[1] 董学松 刘永臻 沈洪令 ZHENG Rui;ZHANG Jixiang;DONG Xuesong;LIU Yongzhen;SHENG Hongling(School of Mechanical-electronic and Vehicle Engineering,Chongqing Jiaotong University,Chongqing 400074,China;Xinli Environmental Technology(Shandong)Co.,LTD,Liaocheng 252022,China)

机构地区：[1]重庆交通大学机电与车辆工程学院,重庆400074 [2]新力环境科技(山东)有限公司,山东聊城252022

出　　处：《探测与控制学报》2024年第5期124-131,共8页Journal of Detection & Control

基　　金：重庆市技术创新与应用发展重点专项(鲁渝科技协作计划项目cstc2020jscx-lyggX0007);重庆市研究生导师团队建设项目(JDDSTD2019007);重庆市研究生联合培养基地(JDLHPYJD2020032)。

摘　　要：针对永磁同步电机(PMSM)转速超调量大、转子位置检测精度低等问题,提出一种BP神经网络智能PID滑模观测器控制策略,将BP神经网络与传统PID控制相结合,利用BP神经网络实现对PID增益的在线调节,实现对永磁同步电机启动、突加负载干扰时稳定控制。采用无位置传感器控制,在永磁同步电机数学模型α-β坐标系中建立了滑模观测器结构,并且在Matlab/Simulink仿真系统中建立了仿真模型进行了仿真分析;从PID参数、电机转速等方面对BP神经网络智能PID控制的有效性进行了评估和仿真验证。通过仿真分析,采用滑模观测器检测转子实际位置与预期位置之间的误差小于7%,在0.3 s之后转子实际位置与预期位置完全重合。采用BP神经网络智能PID控制的永磁同步电机在启动时转速超调量减少了10.6%,在突加负载干扰时减少了1.4%。相比起传统PI控制,提出的BP神经网络智能PID控制能够有效提高PMSM的自适应性及抗干扰能力,并且显著减少了电机在启动及突加负载时超调量。In order to solve the problems of permanent magnet synchronous motor(PMSM)such as large overshoot and low detection precision of rotor position,BP neural network-intelligent PID sliding-mode observer control method was proposed in this paper.BP neural network was combined with traditional PID control and BP neural network was used to regulate PID gain online and control PMSM stably at the time of start and impact load interference.The sliding-mode observer structure was built in the coordinate system of PMSM mathematical model by means of position-sensorless control,and the simulation model was built in MATLAB/Simulink simulation system for simulation analysis;finally,the effectiveness of BP neural network-intelligent PID control was assessed and verified by simulation from aspects of PID parameter,motor speed,etc.Through simulation analysis,the error between the actual position and the expected position of rotor detected with sliding-mode observer was less than 7%,and the actual position exactly coincided with the expected position after 0.3 seconds.The overshoot of PMSM using BP neural network-intelligent PID control reduced by 10.6%at the time of start and reduced by 1.4%at the time of impact load interference.The results showed that compared with traditional PID control,BP neural network-intelligent PID control could greatly improve the self-adaptability and the capacity of resisting interference of PMSM and significantly reduced the overshoot of PMSM at the time of start and impact load.

关 键 词：永磁同步电机 BP神经网络 智能PID 滑模观测器 无位置传感器控制 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]