Vibration Prediction Method Of Railway Vehicle Base On Neural Network

The vehicle vibration is one of the important factors which led to the decrease of vehicle comfort and safety, also it is the main basis for the speed up and limit. Three kinds of vehicle system prediction model are established based on BP, RBF and NARX neural network in this article, which reflecting the relationship between the track irregularity and vehicle vibration. In order to verify the feasibility of the algorithm, a vehicle SIMPACK dynamics model is established and validated based on the measured data, the result shows the SIMPACK dynamics model is with high compatibility compared with the actual vehicle.The validation results based on simulation data from SIMPACK model, show that the NARX neural network prediction output is with high relevance and smaller root mean square error value compared with target output, the neural network model can well predict the change trend of vehicle vibration acceleration. The prediction accuracy of BP and RBF neural network is low, so the genetic algorithm is used to optimize the training process of BP and RBF neural network, and to improve the prediction precision of the models. On the basis of the above algorithms, a vibration prediction virtual experiment system of railway vehicle is designed based on Lab VIEW software, which implements forecasting the vehicle-body vibration acceleration accurately combining with the testing data from track inspection car. The main research contents in this paper are as follows:In this paper, a certain type vehicle dynamics model is set up and verified by comparing the measured vehicle-body vibration data with output data from the model. On this basis, the measured track irregularity data is treated as the model input, then get the vibration response of the vehicle, which aimed to provide the supporting data for the subsequent algorithm validation.Three kinds of vehicle system prediction model are established based on BP, RBF and NARX neural network with the input of track irregularity, and output of vehicle-body vibration acceleration. The traversing method is used to obtain the best neural networkstructure and parameters, and raise the forecast accuracy. The prediction accuracy of BP and RBF neural network is low, so the genetic algorithm is used to optimize the training process of BP and RBF neural network, and to improve the prediction precision of the models. The results show that NARX neural network and optimized BP, RBF neural network, can well predict the vehicle-body vibration acceleration with the track irregularity input, and the predicted value is moderate related to the target value.On the basis of the above algorithms, a vibration prediction virtual experiment system of railway vehicle is designed based on Lab VIEW software, which implements forecasting the vehicle-body vibration acceleration accurately combining with the testing data from GJ-5 track inspection car.