Adaptive neighborhood selection based on locally linear embedding for the degradation index construction of traction motor bearing. (16th August 2021)

• Record Type: Journal Article
• Authors: Zhang, Xingwu; Yu, Xiaolei; Liu, Yilong; Yang, Zhibo; Gong, Baogui; Chen, Xuefeng
• Is Part Of: Measurement science & technology. Volume 32:Number 11(2021)
• Publication Date: 2021-08-16
• Subjects: adaptive neighborhood -- bearing degradation assessment -- degradation index construction -- locally linear embedding; Physical measurements -- Periodicals; Scientific apparatus and instruments -- Periodicals; Equipment and Supplies -- Periodicals; Science -- instrumentation -- Periodicals; Technology -- instrumentation -- Periodicals; Mesures physiques -- Périodiques; Physical measurements; Scientific apparatus and instruments; Periodicals; 502.87
• DOI: 10.1088/1361-6501/ac18a5
• Languages: English
• ISSNs: 0957-0233

Abstract

Abstract: The degradation index construction is significant for bearing degradation assessment which ensures the reliability of machines. Statistical features extracted from vibration signals contain abundant information about the bearing operation state, but not all features have good characterization ability. Therefore, a multi-criteria weighted evaluation criterion is introduced to select features that can properly describe degradation assessment. To construct a more effective and reliable degradation index from the high-dimensional feature set, an adaptive neighborhood selection algorithm based on locally linear embedding (ANS-LLE) is proposed in this paper. The initial neighborhood parameters are determined based on cosine similarity analysis. Then neighborhood parameters are adjusted based on the analysis of sample distribution density and manifold curvature. The effectiveness of the proposed method is validated by an accelerated life experiment and a fault simulation experiment. The results show that the proposed method can effectively describe the bearing degradation process, and ANS-LLE performs better compared with comparison methods.