Sparsity assisted intelligent recognition method for vibration-based machinery health diagnostics

Abstract

Vibration-based health diagnostics technique has shown great potentials to enhance the safety and reliability for many industrial rotary machinery. The emerging sparse representation classification (SRC) paradigm provides a promising tool for intelligent machinery health diagnostics. However, traditional SRC approaches neglect the useful priori information in rotary machinery vibration data, limiting the reconstruction ability and thus restricting their diagnostic accuracy. To address this issue, we present a novel sparsity assisted intelligent recognition (SAIR) methodology for vibration-based machinery health diagnostics. SAIR is constituted by two stages for dictionary design and intelligent recognition. In the dictionary design stage, SAIR exploits the useful priori information of the prediction shift-invariance property, to design class-specific dictionaries via an overlapping segmentation strategy. Additionally, this dictionary design strategy can leverage those local and nonlocal features within data segments. In the intelligent recognition stage, SAIR implements health state recognition by means of the sparsity-based health diagnostic strategy according to minimal sparse approximation errors. Finally, the feasibility and advantage of SAIR have been comprehensively evaluated for planetary gearbox health diagnostics, indicating that SAIR yields an overall diagnostic accuracy of 99.72%. Besides, comparative studies demonstrate that SAIR outperforms the advanced mainstream methods with better diagnostic accuracy and lower computation costs for vibration-based machinery health diagnostics.

Clustering Federated Learning for Bearing Fault Diagnosis in Aerospace Applications with a Self-Attention Mechanism

“…Various research on essential diagnosis issues, such as deep learning methods [3,4], knowledge transfer [5][6][7][8][9], fault decoupling and detection [10][11][12], imbalance data augmentation, and model generalization [13][14][15][16], have been carried out. For example, Syed Muhammad Tayyab et al [17] used machine learning through optimal feature extraction and selection for intelligent fault diagnosis of machine elements.…”