基于神经网络的无刷同步电机控制系统的故障诊断研究

设计了一套无刷同步电机控制器的故障诊断系统,利用实际的非线性控制系统训练神经网络状态观测器,根据系统实际输出与神经网络观测器输出之间的残差来判别和检测实际控制系统的故障,针对系统控制器、电流和速度传感器故障进行仿真实验故障诊断研究。仿真结果表明该方法能有效抑制噪声,并快速准确地确定故障发生的时间、位置、大小以及故障的类型,具有很强的鲁棒性,且适用性和应用价值强。

Fault diagnosis of brushless synchronous motor based on neural network

A fault-diagnosis method for controller of brushless permanent magnet synchronous motor is analyzed,the observer based neural network is trained by real nonlinear control system.The fault-detection for controller,sensor of current and speed are discussed by observer of neural network.The simulation results show that the method can isolate any fault of the sensors and controller in any input types and normal work frequency,that neural networks are the powerful tools to approximatis almost every non-linear system.The diagnosis method has very strong suitability and practical value.