导航

ACTA AERONAUTICAET ASTRONAUTICA SINICA ›› 2023, Vol. 44 ›› Issue (4): 426887-426887.doi: 10.7527/S1000-6893.2022.26887

• Material Engineering and Mechanical Manufacturing • Previous Articles    

Parallel sparse filtering for fault diagnosis under bearing acoustic signal

Jinrui WANG1, Shanshan JI1, Zongzhen ZHANG1,2(), Zhenyun CHU1, Baokun HAN1, Huaiqian BAO1   

  1. 1.College of Mechanical and Electronic Engineering,Shandong University of Science and Technology,Qingdao  266590,China
    2.College of Energy and Power Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing  210016,China
  • Received:2021-12-31 Revised:2022-01-14 Accepted:2022-02-22 Online:2022-04-01 Published:2022-03-30
  • Contact: Zongzhen ZHANG E-mail:zhzz18@126.com
  • Supported by:
    National Natural Science Foundation of China(52005303);Natural Science Foundation of Shandong Province(ZR2020QE157)

Abstract:

Vibration signal is commonly used in fault monitoring of aeroengine. However, the arrangement of vibration sensors is increasingly strict due to the complex structure of aeroengine. Acoustic signal has attracted extensive attention in intelligent fault diagnosis of bearings because of its advantages of non-contact, easy arrangement and low cost. However, traditional bearing fault diagnosis methods cannot achieve accurate feature extraction due to the strong ambient noise in the aeroengine acoustic signal. Therefore, to realize intelligent fault diagnosis under bearing acoustic signals, it is particularly important to study effective feature extraction methods. Sparse representation is a research hotspot in intelligent fault diagnosis, which shows great power in sparse feature extraction. Effective sparse feature extraction of acoustic signals under strong noise can provide a solution for non-contact fault diagnosis of bearings. In this paper, a novel fault diagnosis method based on parallel sparse filtering is proposed to realize sparse feature extraction of bearing acoustic signals. Specially, parallel sparse filtering achieves further sparse feature exaction by adding another normalization direction on the basis of traditional sparse filtering, and then the weight normalization method is used to constrain the weight matrix obtained by network training. Finally, the superiority of the proposed method is validated by simulation and experimental dataset. The results show that parallel sparse filtering can realize effective feature extraction and accurate classification of bearing acoustic signals, and can be used for mechanical intelligent fault diagnosis under acoustic signals.

Key words: bearing, fault diagnosis, acoustic signal, parallel sparse filtering, feature extraction

CLC Number: