小波包和改进 Elm an 神经网络相融合的异步电动机滚动轴承的故障诊断

摘要

针对单一的信号处理方法不易准确诊断出异步电动机滚动轴承故障的问题，提出小波包和改进Elman神经网络相融合的诊断方法。利用小波包对采集的4种不同故障信号的数据进行去噪、分解和重构，有效地提取出不同故障类型的能量特征，并通过引入自反馈因子β构建改进的Elman神经网络。实验诊断结果表明：改进前后的Elman神经网络均能实现对异步电动机滚动轴承的故障诊断，但就诊断时间和精度而言，后者比前者具有更高的诊断效率和准确度。%As for the limitations of the single signal processing method is not easy to accurately diagnose fault of asynchronous mo -tor rolling bearing ,the diagnosis method of the wavelet packet and improved Elman neural network are fused is proposed .The ac-quisition data of four different fault signals are denoised ,decomposed and reconstructed using the wavelet packet ,the energy char-acteristics of different fault types is extracted in effect .By introducing the βof a feedback factor build neural network of improved Elman.Experimental results show that both of improved and unimproved Elman neural network can realize the fault diagnosis of a -synchronous motor rolling bearing ,but as for the diagnosis time and accuracy ,the latter is better than the former in the diagnostic efficiency and accuracy .