Data-driven Train Running Visualization

The monitoring and evaluation of the three major dynamic performance indicators for running train: the motion stability,running stability,and curving performance are usually based on line experiments or shaking table experiments.By processing a large amount of experimental data,the dynamic performance index of the train can be analyzed.With the rapid development of computer simulation and visualization technologies,the simulation of trains under various operating conditions has become an effective alternative method to analyze and evaluate the stability and safety of trains.However,the traditional train dynamics simulation software can provide a large amount of simulation data to analyze and study the running status of the train under various conditions,but it provides limited functions to model and visualize the train's real operating environment(terrain,weather,light,road conditions,etc.).In the other hand,the mainstream visualization system of train running only focuses on the visualization of the train operation,ignoring the analysis of the dynamic performance of the train.In view of this situation,this paper proposes a data-driven model to predict and calculate the stationarity index value of trains based on multi-body dynamics simulation results.In addition,a 3D visualization system of train running was developed by using a 3D visualization engine,UNIGINE.The function of real-time calculation and display of the stationarity index was integrated into the system.To model the 3D terrain rapidly,we developed a plug-in for UNIGINE which can automatically model large-scale 3D terrain based on 3D GIS(Geographic Information System)data.With a few manual interventions,the entire 3D scene of train operations can be built conveniently.After the model was constructed,the model was simply optimized.Using the multi-body dynamics software SIMPACK,the dynamic behaviors of the train under different operating conditions was simulated.Based on the simulation results,the stationarity of the running train was evaluated.For the case of the train model running at a fixed speed which under different levels of U.S track spectrums,the prediction and calculation results of the BP neural network and the least-squares polynomial fitting method are analyzed and compared;and for the case of the train running at different speeds which under the fixed track spectrum,a least squares linear regression model was established to predict the vertical stability of the train.Finally,with the aid of the 3D visualization engine UNIGINE,3D visualization of the dynamic process of the train running on a straight track is realized.