Predicting Vertical Acceleration of Railway Wagons Using Regression Algorithms

The performance of rail vehicles running on railway tracks is governed by the dynamic behaviors of railway bogies, particularly in cases of lateral instability and track irregularities. To ensure reliable, safe, and secure operation of railway systems, it is desirable to adopt intelligent monitoring systems for railway wagons. In this paper, a forecasting model is developed to investigate the vertical-acceleration behavior of railway wagons that are attached to a moving locomotive using modern machine-learning techniques. Both front- and rear-body vertical-acceleration conditions are predicted using popular regression algorithms. Different types of models can be built using a uniform platform to evaluate their performance. The estimation techniques' performance has been measured using a set of attributes' correlation coefficient (CC), root mean square error (RMSE), mean absolute error (MAE), root relative squared error (RRSE), relative absolute error (RAE), and computational complexity for each of the algorithms. Statistical hypothesis analysis is applied to determine the most suitable regression algorithm for this application. Finally, spectral analysis of the front- and rear-body vertical condition is produced from the predicted data using the fast Fourier transform (FFT) and is used to generate precautionary signals and system status that can be used by a locomotive driver for necessary actions.