一种交流电机故障诊断中的特征提取与强化

A Feature Enhancement and Extraction Method for Motor Failure Diagnosis

为提高电机故障诊断的准确率和有效性,提出了一种故障特征提取与强化的新方法.即在对所采集的交流电机振动加速度信号进行数据预处理之后,用盲源分离方法进行独立振动源的分离,然后采用小波包分析方法进行特征提取,并进行特征频带的简化及特征强化处理,特征强化后的数据作为交流电机故障诊断模型的输入.该方法通过对振动加速度信号进行分离,能够分离出混合信号中的独立振动源,提高了故障特征提取的正确率和准确度;通过对特征频带化简,减少了故障诊断模型的输入,进而简化了模型的结构;特征强化使模型能够更有效地识别故障状态.

In order to improve the accuracy and effectiveness of motor fault diagnosis,a new method of fault feature enhancement and extraction is proposed.After data preprocessing the sampled AC motor vibration acceleration signal,an independent vibration source separation with blind source separation method is executed,then the feature extraction with wavelet packet analysis method and feature frequency band simplifying and feature enhancement are conducted,and the feature enhanced data are used as the input to the AC motor fault diagnosis model.With this method,the independent vibration source in the mixed signal can be separated by separating vibration acceleration,and improved the accuracy of fault feature extraction.With the feature band simplification,the input to the fault diagnosis model can be reduced with simplified model structure.Feature enhancement model will help to identify and predict failure more effectively.