Digital Twin for wear degradation of sliding bearing based on PFENN

The emergence of "Digital Twin"has raised higher demands for the health management of industrial equipment. Sliding bearings, as crucial supporting components in rotating machinery, may wear under radial loads, posing potential risks to the safety and stability of the system. However, The wear profile is usually challenging to measure online, making it difficult to present the real-time wear status of the bearing. To address this issue, this paper proposes a novel framework of digital twin by combining physics -driven with data -driven model in parallel. Within this framework, we establish a parallel hybrid model of finite element and deep neural network (PFENN), used for monitoring and visualizing the real-time wear profile of sliding bearing. In the offline phase, PFENN gets the wear profile through numerical calculations of finite element model and constructs the space of bearing wear states; during the online phase, PFENN captures the bearing's vibration signal and obtains the maximum wear depth through deep neural network. Subsequently, mapping the wear profile in the wear data space based on the maximum wear depth and visually presenting. Additionally, we employ relevance vector regression with a sliding window to fit the bearing degradation curve for predicting the remaining useful life. Experimental validation confirmed PFENN's 93% profile accuracy and the capability for real-time monitoring and prediction, meeting the demands for high-fidelity digital twinning and real-time performance. This research also introduces new methodologies and perspectives for industrial equipment health management.