Wear performance prediction of a novel bearing based on neural network algorithms: Theory and experiment

The paper focuses on the complex wear performance of water-lubricated bearings during start-up phase. The boundary conditions reflecting the actual operations are set. Instantaneous wear model is constructed. Influences of diverse operating and structural parameters on performance are explored. The pressure distributions and wear characteristics in various rotational speeds, coefficient of friction, length-diameter ratios, materials are studied. The wear-depth predict model with BP neural network algorithm is built. The nonlinear relationships between multiple factors and wear depth are revealed. The accurate prediction of wear depth is realized. Results indicate increasing load leads obvious growth of contact pressure. Contact pressure is an important factor affecting wear state. The paper provides the optimum design and life evaluation technical support for such bearings.