基于改进遗传小波网络的电机故障诊断研究  

Fault Diagnosis Study for Motor Based on Improved Genetic Algorithm Wavelet Neural Network

作　　者：吕永健[1] 李子龙[1] 张洪林[1] 

机构地区：[1]空军工程大学,陕西西安710038

出　　处：《微特电机》2012年第4期19-21,37,共4页Small & Special Electrical Machines

摘　　要：用频谱分析方法提取了无刷直流电动机的正常工作状态和几种常见的故障(位置传感器一路故障、A相绕组断路故障和驱动开关断路故障)时的特征信号,进行了诊断算法研究,提出了用改进遗传算法优化小波神经网络参数的调整过程,并用改进遗传小波神经网络对无刷直流电动机进行故障诊断。仿真结果表明,与经典遗传小波神经网络、小波神经网络和BP神经网络等方法进行比较,该方法在无刷直流电动机故障诊断中具有更快的收敛速度和更高的诊断精度。Spectral analysis was adopted to pick up characteristic signals when brushless DC motor was normal or in several general faults such as state with one breaking position sensor,one breaking phase winding or one breaking drive switch.A fault diagnosis method to supervise and diagnose faults in brushless DC motor was proposed based on a wavelet neural network with improved genetic algorithm,which was applied to find global optimization values for the network weights.The method was more effective and feasible,compared with those methods of a wavelet neural network with classical genetic algorithm,a wavelet neural network and BP neural network in the instance simulation of the fault diagnosis for brushless DC motor.

关 键 词：无刷直流电动机 故障诊断 改进遗传算法 小波神经网络 

分 类 号：TM33[电气工程—电机]