Rolling Bearing Fault Prediction Based On Feature Fusion Of HI Curves And GRU Networks

The study of rolling bearing fault prediction is significant in the context of today's increasingly sophisticated machinery, where traditional prediction methods based on physical models fail to achieve prediction results due to the complexity of the mechanical structure. In this paper, a fault prediction model based on health indicators (HI) created by feature fusion algorithm combined with gated recurrent unit (GRU) network is proposed to build a new HI with root mean square, peak, root mean square frequency and frequency center of gravity for feature fusion, and GRU network is used as the core to build the prediction model of health indicators. According to the prediction results, the proposed HI combined with GRU network has the highest prediction accuracy in contrast to techniques like support vector machine and long and short-term memory, which can more precisely predict the degradation trend of rolling bearings, thus effectively predicting the fault of rolling bearings and reducing the loss caused by the occurrence of fault.