Fault diagnosis of rotating machinery in a noisy environment based on modified general normalized sparse filtering

Intelligent fault diagnosis for mechanical condition monitoring has achieved a great deal of success in recent years, but most of the research is carried out in experimental environments. Vibration signals collected in real scenarios usually have strong noise interference, which significantly reduces fault classification capability and seriously affects the accuracy and robustness of classification. This paper proposes modified general normalized sparse filtering (MGNSF) with strong noise adaptability for rotating machinery fault diagnosis without any time-consuming denoising preprocessing, in which generalized normalization, weight and feature normalization, and the Hankel matrix. Diagnostic performance is studied with the change of normalization parameters and signal noise ratio. Weight and feature normalization can improve the distribution between features. The proposed algorithm is validated using two rolling bearing datasets. The experimental results show that MGNSF can extract the features of a faulty bearing under stronger noise interference and has strong noise adaptability.