神经网络优化非线性磁链的双三相PMSM无感控制  

Sensorless Control of Dual Three-Phase PMSM Based on Nonlinear Flux Observer Optimized by Neural Network

作　　者：赵化勇 田伟 吉敬华[1] ZHAO Huayong;TIAN Wei;JI Jinghua(School of Electrical and Information Engineering,Jiangsu University,Zhenjiang 212013,China)

机构地区：[1]江苏大学电气信息工程学院,镇江212013

出　　处：《组合机床与自动化加工技术》2025年第4期129-132,139,共5页Modular Machine Tool & Automatic Manufacturing Technique

摘　　要：针对中低速情况下双三相永磁同步电机无传感器控制精度低,非线性磁链观测器在静止和低速时稳定性差的问题,提出一种结合BP神经网络的离散型非线性磁链观测器。首先,结合电流方程构建非线性磁链观测器并利用欧拉离散法进行离散化;其次,采用BP神经网络优化非线性磁链的固定增益,实现增益在线调节,同时提出自适应高频信号注入方法,解决了非线性磁链观测器受固定增益值限制观测误差大和低速情况下稳定性差的问题;最后,在MATLAB环境下搭建了仿真模型验证提出的算法,并且仿真结果显示新型观测器的位置误差减小了37.5%以上,收敛速度提升了50%以上,有效地抑制系统抖振,具有更强的鲁棒性。Aiming at the problems of low sensorless control accuracy and poor stability of nonlinear flux observer at static and low speed,a discrete nonlinear flux observer combined with BP neural network was proposed.Firstly,the nonlinear flux observer is constructed by combining the current equation and discretized by the Euler discretization method.Secondly,BP neural network is used to optimize the fixed gain of the nonlinear flux observer to achieve online gain adjustment.At the same time,an adaptive high frequency signal injection method is proposed to solve the problem that the nonlinear flux observer is limited by the fixed gain value and the stability is poor at low speed.Finally,a simulation model is built in MATLAB environment to verify the proposed algorithm,and the simulation results show that the position error of the new observer is reduced by more than 37.5%,the convergence speed is increased by more than 50%,and the system chattering is effectively suppressed with stronger robustness.

关 键 词：双三相永磁同步电机 矢量控制 离散型非线性磁链观测器 BP神经网络 高频信号注入 

分 类 号：TH165[机械工程—机械制造及自动化] TG659[金属学及工艺—金属切削加工及机床]