Tribological properties prediction of brake lining for automobiles based on BP neural network

By many tribological experiments of brake lining for automobiles, the original experimental data were firstly obtained, which contains the influencing rules of braking conditions on tribological performance. Based on the artificial neural network technology and the experimental data specimens, the BP neural network model was established to predict the tribological properties. Three parameters of braking conditions (braking pressure, sliding velocity and surface temperature) were selected as input vectors, and two parameters of tribological performance (friction coefficient and wear rate) were selected as output vectors. By contrast of prediction values and experimental results, it is found that the neural network can predict properly the influencing rules of braking conditions on tribological performance. What is more, the neural network has quite favorable ability for predicting of friction coefficient. While it has bad ability for predicting of wear rate, especially when the pressure, velocity and temperature are high. As a whole, this paper has proved that it is feasible and valuable to use neural network for predicting tribological properties of friction materials.