Sound-Insulation Prediction for High-Speed Train Walls Based on Neural Network Learning

In this paper, a method and model of sound insulation prediction for the side wall structure of a high-speed train based on back propagation neural network is proposed. Firstly, the components of high-speed train side walls are classified and analyzed. Based on the measured results and existing literature, the main factors affecting the sound insulation characteristics of side walls are listed to form a feature set. Then, correlation analysis and redundancy analysis were carried out for the above feature sets, and the optimal feature subset was determined based on the maximum correlation—minimum redundancy (M-RMR) criterion. Finally, the BP neural network model is established and trained with the sound insulation volume of side wall structure as the objective function. The results show that compared with the traditional FEM model, the BP neural network model significantly improves the computational efficiency and accuracy.