As critical fundamental components in high-end equipment, angular contact ball bearings undergo service performance degradation that directly impacts the operational accuracy and reliability of the entire system. Traditional numerical methods suffer from low computational efficiency, whereas simplified theoretical frameworks cannot accurately capture the dynamic coupling mechanisms between bearing assembly parameters and service performance. This limitation leads to an inevitable trade-off between efficiency and accuracy in predicting dynamic responses. To address this issue, we propose a Backpro Pagation (BP) neural network–finite element modeling framework. First, a dynamic model of the angular contact ball bearing is constructed. Second, representative assembly process parameters are selected as inputs, and performance indicators are defined as outputs, with the dataset generated via finite element simulations. Third, a surrogate model is developed to predict bearing service performance. Finally, the effectiveness and accuracy of the surrogate model are validated experimentally, thereby providing a solid theoretical foundation and technical support for condition-based maintenance and intelligent health management of bearings.

bearings, surrogate model , neural network , finite element analysis

暂无