Aiming at the challenge of weak compound fault diagnosis of rotating machinery,a novel method named intrinsic component filtering(ICF)is proposed for signature detection and separation under noisy envi-ronments norms of the rows and norms of the columns are used to achieve the sparse distribution in each sample and consistency among samples,respectively.Optimum filters are learned through minimizing the objective function.First,Hankel training matrix of the input signal is constructed,and the convolution process is simu-lated by the product of the weight matrix and Hankel matrix.Then,ICF is used to learn the feature matrix.The final optimum filters are determined through the Kurtosis of the trained filters.Finally,we can diagnose the fault condition using the extracted features and the corresponding envelope spectral.The simulated and experimental fault data are used to validate the performance of the proposed method.The results confirm that the proposed method can separate the weak fault components and guarantee strong robustness under strong noisy environment without any prior experience.
维普
万方数据