基于BP神经网络和旋转高频电压注入法的永磁同步电机转子位置自检测.pdfVIP

基于BP 神经网络和旋转高频电压注入法的 永磁同步电机转子位置自检测 1 韦 玮，贾利斌，薛凤举，王 蕾 （北京航天控制仪器研究所，北京100854） 永磁同步电机的运动是一个强耦合的非线性动态控制系统，存在大的干扰噪声以及参数的变化，在定子电 摘要： 流测量过程中存在测量误差，采用高频电压信号注入法进行永磁同步电机位置自检测时传统线性控制理论很难达 到要求。设计了一种基于BP神经网络整定的PID观测器,对旋转高频电压信号注入法所得的高频载波电流进行解调 后，送入观测器进行转子位置估测。仿真和实验结果表明,当系统存在系统噪声、测量噪声及参数变化时,基于BP 神经网络整定的PID位置观测器能够有效去除干扰和系统参数变化带来的影响,对转子位置进行比较准确的估计。 关键词：永磁同步电机；无位置传感器；旋转高频信号注入法；BP神经网络 Study on Rotor Position Self-sensing of a PMSM Based on Rotating High-frequency Injection and BP Neural Networks WEI Wei, JIA Libin, XUE Fengju, WANG Lei (Beijing Aerospace Control Instrument Research Institute, Beijing 100854) Abstrac ：The control of PMSM is a strong coupling nonlinear dynamic system, it would be noisy in the parameter detecting during the control course, while the motor ’s parameter is changing. The traditional linear control theory fails to meet the system requirement. A BP neural networks PID estimator, which is used for rotor position self-sensing under high-frequency voltage injection, is proposed in this paper. By utilizing rotating high-frequency injection to realize rotor position self-sensing of PMSM, after demodulating the produced high-frequency carrier current, it will be sent to BP neural networks PID estimator, so that to eliminate the noise, and obtain the rotor position accurately. The simulated experiment result shows that under the noisy circumstances and motor ’s parameter changing, the rotating high-frequency injection based on BP neural networks PID estimator could track the rotor position accurately. Keywords：PMSM，position sensorless，high-f requency inj ection ，BP neural networks 1 引言 永磁同步电机具有体积小、效率高、可靠性好及对环境的适应性强等诸多优点，在