为预测某一成分的增强PA6基摩擦材料在特定摩擦条件下的摩擦磨损性能,建立了基于BP和RBF神经网络的混合神经网络,并利用自适应遗传算法优化网络参数。模型经训练后进行实际预测,预测值与实测值的相关系数在0.99以上,并描绘出了摩擦因数和磨损率分布关于材料组成或测试条件的三维分布图,结果表明该模型有良好的预测精度和指导意义。 In order to predict the friction and wear properties of a PA6-based friction material under certain friction conditions, a hybrid neural network based on BP neural network and RBF neural network was established and the network parameters were optimized by using adaptive genetic algorithm. The model was predicted after training and the correlation coefficient between the predicted value and the measured value was above 0.99, and the distribution of friction coefficient and wear rate was plotted on the three-dimensional distribution of material composition or test conditions. The results show that the model has good prediction accuracy And instructive.