Parallel sparse filtering for intelligent fault diagnosis using acoustic signal processing

Acoustic signals have attracted considerable attention in mechanical fault diagnosis because of their advantages in non-invasive technique, instant measurement and low cost. However, traditional fault diagnosis methods could not achieve accurate feature extraction because of the strong noise environment of acoustic signals. In view of this, this study aims to provide a method that could accurately extract effectiveness features under noisy environment. Sparse representation is a research hotspot in intelligent fault diagnosis and has shown great power in feature extraction. In this paper, a novel fault diagnosis method based on parallel sparse filtering is presented to achieve sparse feature extraction from acoustic signals. Specially, parallel sparse filtering achieves sparse feature exaction by adding another normalization direction based on sparse filtering, and the derivation of parallel sparse filtering is also presented in detail. Then Z-score normalization is used to activate the training and testing data in fault classification process. The superiority of the proposed method is validated by simulated and experimental data. The results show that PSF is a promising sparse feature extraction method that can be used for mechanical fault diagnosis under acoustic signals. (c) 2021 Elsevier B.V. All rights reserved.