A performance degradation assessment method using one-dimensional sparse representation self-learning dictionary

Abstract

Rotating machinery is a widely-used component of factories and enterprises. Unquestionably, rolling bearing is the backbone of rotating machinery and its performance degradation assessment (PDA) is a vital part of condition-based intelligent health management. Therefore, a PDA method based on one-dimensional sparse representation self-learning dictionary is presented to assess the degradation trend of rolling-element bearings. We proposed a novel feature extraction method named one-dimensional K-singular value decomposition to extract the sensitive feature from run-to-failure data. Then, an improved support vector data description is adopted to establish the PDA model, and a novel degradation indicator is obtained. Subsequently, the proposed method adaptively tracks the development of deterioration and identifies the occurrence of an incipient fault. Experimental results on the simulation signal and XJTU-SY bearing datasets verify the effectiveness and outperform the conventional methods.