Multiscale local features learning based on BP neural network for rolling bearing intelligent fault diagnosis

Traditional intelligent fault diagnosis techniques based on artificially selected features fail to make the most of the raw data information, and are short of the capabilities of feature self-learning. Moreover, the most informative and distinguished parts of the different faults signals only account for a small portion in the time domain and frequency domain signals. Therefore, in order to learn the discriminative features from the raw data adaptively, this paper proposes a multiscale local feature learning method based on back-propagation neural network (BPNN) for rolling bearings fault diagnosis. Based on the local characteristics of the fault features in the time domain and the frequency domain, the BPNN is used to locally learn meaningful and dissimilar features from signals of different scales, thus improving the fault diagnosis accuracy. Two sets of rolling bearing datasets are adopted to verify the validity and superiority of the proposed method by comparing with other methods. (C) 2019 Elsevier Ltd. All rights reserved.