基于小波包变换和奇异值分解的风机故障诊断研究

Wavelet packet transform and singular value decomposition based fault diagnosis of fans

为了通过振动信号准确识别风机的工作状态,提出了利用小波包变换和奇异值分解提取振动信号特征的方法。该方法首先对振动信号进行小波包分解,用分解系数的重构信号构建特征矩阵,然后对此矩阵进行奇异值分解得到其奇异值特征向量,并统计出特征向量的最大值、最小值和均值作为识别风机机械故障的特征参数,最后计算5个测点测量所得振动信号的特征参数,并将其融合得到风机故障小波包奇异值特征向量,再采用动量法和学习速率自适应的改进BP神经网络进行故障诊断。结果表明,该方法能有效地诊断出风机机械故障的类别、程度和发生部位。

Aimed at identifying the fans' working status accurately,a method using wavelet packet transform and singular value decomposition to extract the vibration signal characteristics was proposed.This method conducted wavelet packet decomposition on the vibration signal,and applied the reconstruction signals of wavelet packet decomposition coefficients to build the characteristic matrix.Then the singular value feature vector of this matrix was obtained with utilization of the singular value decomposition.Through the statistics analysis,the maximum,minimum and mean value of the singular value feature vector was taken as characteristic parameters to recognize the fan mechanical fault.Finally,a fan fault feature vector was put forward in order to diagnose the mechanical failure of fans accurately,which was the integration of the vibration signal wavelet packet singular value features of the five measuring points in the same instantaneous running state,and then the fault diagnosis of fan was achieved by applying the improved back propagation(BP)neural network.The diagnostic results show that this method can diagnose the category,severity and location of the fan mechanical failures effectively.