基于Gammatone倒谱系数的异步电机故障诊断方法研究

Research on Fault Diagnosis Method for Asynchronous Motor Based on Gammatone Cepstrum Coefficient

异步电机运行声音包含着许多重要信息,对声学特征的有效提取结果可用于故障检测。因此,提出一种基于可听声信号的异步电机故障诊断方法。以异步电机为实验对象,利用Gammatone滤波器提取一种正常工况和两种异常工况的声音信号GFCC特征向量;采用卷积神经网络对可视化后的特征量进行分类识别。结果表明:相较于利用梅尔滤波器提取的两种特征量,利用GFCC提取的特征量具有更高的目标识别率。

The running sound of an asynchronous motor contains a lot of important information,and the effective extraction of acoustic features can be used for fault detection.Therefore,a fault diagnosis method of asynchronous motor based on audible sound signal was proposed.Taking the asynchronous motor as the experimental object,the Gammatone filter was used to extract the GFCC feature vector of the sound signal in a normal working condition and two abnormal working conditions;a convolutional neural network was used to classify and recognize the visualized feature quantity.The results show that compared with the two feature quantities extracted by using the Meir filter,the feature quantities extracted by using the GFCC have higher target recognition rate.