A new raw signal fusion method using reweighted VMD for early crack fault diagnosis at spline tooth of clutch friction disc

Abstract

Spline tooth root crack is a kind of critical faults for clutch friction disc. Most traditional crack diagnosis methods based on single-sensor or fusion may be degraded due to incomplete fault information or low signal-to-noise ratio (SNR). Aiming at enhancing the ratio of fault-related information in the fusion signal, a new reweighted variational mode decomposition (VMD) multi-point fusion method is proposed. With the proposed method, each measurement point vibration acceleration signal is decomposed by VMD firstly, and the intrinsic mode functions (IMFs) containing different sensitivities information are obtained. Then, correlation energy fluctuation (CEF) and confidence distance are utilized to evaluate the IMFs and raw signal respectively. A new dual-weight strategy is proposed to restructure the IMFs. Finally, convolutional neural network (CNN) is used to identify spline tooth early crack faults. Experimental results demonstrate that the proposed method has greater fault recognition rate than single point and other fusion methods.