Assessment of vehicle comprehensive maintenance capability based on RBF neural network

Abstract

To solve the problems of experience-based and subjective-based methods of capability assessment on vehicle maintenance, a RBF neural network model is established, which takes the 7 key factors of vehicle maintenance capability as input, and takes the comprehensive maintenance capability as output. The self-organization learning method and least-squares method are used to determine the weight parameters of RBF neural network, based on which the comprehensive maintenance capability method of vehicle is designed. Simulation results show that the final error of the trained RBF neural network is 0.0073, and the percentage errors of neural network calculated based on test samples are less than 5%, which indicates that the established RBF neural network could assess the comprehensive maintenance capability method of vehicle effectively, which means the neural network is more practical.