开关磁阻电机神经网络无位置传感器控制

Position sensorless control of SRM using neural network

针对现有开关磁阻电机(SRM)的转子位置传感器使得系统成本和复杂度提高、坚固性和可靠性降低的问题,研究了SRM无位置传感器DSP控制实现。建立了开关磁阻电机位置检测神经网络模型,并给出了提出对象的学习算法和训练步骤。采用TMS320F2812 DSP实现神经网络在线训练算法,开发完成了一台15kW三相12/8极无位置传感器SRD样机。实验结果表明,无位置传感器SRD具有较好的动态特性和较高精确度,系统最大位置检测误差≤2°。

As position sensor of switched reluctance motor（SRM） increases system cost and complexity,while reduces robustness and reliability,sensorless control of SRM based on DSP was proposed.A neural network of rotor position estimation for SRM sensorless drive was set up.On-line learning algorithms and training steps were also given.Neural network on-line training algorithm was achieved by TMS320F2812 DSP,and 15kW three-phase 12/8 pole sensorless SRD was set up.Experimental results show that the system has a good dynamic performance and high accuracy position detection with maximum error less than 2°.