Prediction and Factor Analysis for Friction and Wear Performance of Brake Disk

In order to obtain friction and wear performance of different brakes in different conditions with less test data, back propagation artificial neural network model has been established by some physical parameters and working conditions to train and predict friction and wear performance of carbon brake disk. The predicted values for training and investigating are accuracy in comparison with the real test data, and factors to influence brake performance have been quantitatively analyzed by principal component analysis. The result shows that heat-sinking capability and working condition might be the primary cause for brake friction and wear difference, and the methods above could be applied to friction and wear performance prediction and factor analysis in engineering practice.