Multivariate Analysis (MA) of Train Swaying Dynamics Based on Wheel Parameters and Operational Speed

Abstract

The aim of this research is to isolate the effect of two basic mechanical degradation mechanisms in the wheels, namely the thickness of the flanges and the diameter of the wheels, on the overall train swaying behavior, as well as to quantify the influence of the kinematic operational speed of the train. The study was based on historical maintenance data sets as well as on dynamic field measurements of 28 different observations. Multivariate analysis, with Multiple Linear Regression (MLR) model, was used to process data. The model was validated to ensure reliability and to normalize any prediction errors with Mean Squared Error (MSE) which yielded a very low error of 0.0017. The statistical analysis gave a coefficient of determination of R² = 0.201 and the individual P-values for flange thickness (0.456), wheel diameter (0.142) and operational speed (0.087) were all higher than 0.05. The results clearly demonstrate that wheel wear and the current operating speeds are not the main contributor towards train swaying in this operational envelope. A mathematical validation of the strong intrinsic stability of the bogie system is obtained for this finding. Finally, future mitigation strategies will need to move beyond wheel reprofiling and focus on the higher-order dynamics that require investigation.