Using the Al₂O₃ ceramic as friction pair，the effects of single influence factor(temperature, load or speed) on the friction factor of MoSi₂ material were studied by high temperature friction and wear testing machine. The wear rate predication of MoSi₂ material under interaction among temperature, load and speed were also studied. The adaptive learning rate and with additional secondary momentum BP neural networks prediction model were studied. The network training process was also given. Network prediction and the actual testing results show that the improved BP neural network has higher prediction accuracy and the prediction error was controlled less than 3%. Network prediction can meet the predicting needs of molybdenum disilicide under the complex conditions.

MoSi₂, friction and wear properties, wear rate, neural network prediction