樊薇学术报告：Separated Sparse Representation Model for Bearing Fault Detection
题 目：Sparse Representation Model for Bearing Fault Detection
Vibration signal from a rolling bearing with localized faults always contains periodic transients and background noise. This talk will present a novel fault detection technique by utilizing the sparsity of the transients. The separated sparse representation (SSR) model with a tunable separation time parameter is constructed. In the implementation of the model, a B-spline dictionary is adopted to represent the transient due to its inherent ability to model sparsity and its impressive flexibility, and then the model is solved by split augmented Lagrangian shrinkage algorithm (SALSA). The power value calculated by the reconstructed signal will reach the maximum when the separation time parameter is the same as the true fault period, which is proposed as a criterion to detect the fault period. The performance of the proposed method is compared with conventional methods in the simulation studies. Finally, case studies are used to illustrate the effectiveness of the proposed methodology in fault period detection.
Wei Fan received her bachelor’s degree and master’s degree both from Soochow University. She is currently working towards the Ph.D. degree in Department of Systems Engineering and Engineering Management at City University of Hong Kong. Her research interests focus on statistical process control, signal processing and machinery fault diagnosis.