复励式无刷直流电机故障诊断研究

Fault Diagnosis Research for Hybrid Excitation BLDCM

中文关键词:  复励式无刷直流电机  小波分析  神经网络  故障诊断

英文关键词:Hybrid excitation brushless DC motor  wavelet analysis  neural network  fault diagnosis

中文摘要:

      针对复励式无刷直流电机弱磁控制的故障信号特点，选择小波神经网络作为电机故障诊断方法。根据电机的故障树，确定了电流作为其故障诊断信号。以最常见的绕组短路和开路作为研究对象，通过对不同小波基函数的对比分析，选择coif5作为小波基函数。利用Mallat算法对典型电机故障信号进行了检测,采用第2层分解时的高频系数d₂作为特征值，采集了多组故障信号特征向量并进行了归一化处理。诊断结果表明,小波神经网络能准确地识别故障信息。

英文摘要:

      A fault diagnosis model was established based on wavelet neural network,according to the fault signal characteristics of hybrid excitation brushless DC motor with flux weakening operation.Current was determined as the fault diagnosis signal due to fault tree.Winding open and short circuit was studied.Coif5 was selected as the wavelet basis function after comparison among different wavelet base functions.The typical motor fault signal was detected through Mallat algorithm.High frequency coefficients d₂,which obtained from second layer decomposition,was used as eigenvalue of winding open and short circuit.Experiment validates that the model is accurate and reliable.