A fault diagnosis method for rolling bearings based on VGG16-ELM

Abstract

As critical components of rotating machinery, the operational health of rolling bearings is a key determinant of the safety and reliability of the entire mechanical system. To overcome the drawbacks of conventional fault diagnosis methods for rolling bearings, such as reliance on manual feature extraction and the weak feature mining capability of shallow models, this paper presents an intelligent fault diagnosis framework based on a VGG16 network fused with an Extreme Learning Machine (ELM). First, the method begins by converting one-dimensional bearing vibration time-series signals into two-dimensional time-frequency images using Continuous Wavelet Transform (CWT), thereby transforming the fault diagnosis task into an image recognition problem. Second, the VGG16 network is utilized to automatically extract deep texture and time-frequency correlation features from these images, removing the requirement for manual feature design. Finally, the high-dimensional feature vectors output by VGG16 are fed into the ELM to perform the fault classification and identification. Experimental results demonstrate that the VGG16-ELM model possesses superior feature extraction capabilities and generalization performance, providing a novel and feasible solution for intelligent fault diagnosis of rolling bearings in industrial field applications.