Abstract
The paper introduces the new method to estimate vertical gauge irregularity from accelerations of the bogie using ICA (Independent Component Analysis) and examines the performance through the numerical simulation of a travelling bogie. The authors propose to apply ICA to monitor the railway system. It is a method utilized mainly in the field of biophysics to identify the signal sources from the multi data sets of the mixed signals. Only the vertical and pitch motion of the bogie are considered and the dynamics of the car body is neglected. To simulate the sensor noises, the white noises are added to the accelerations. The velocities and the displacements are calculated by integrating and double integrating the observed accelerations, respectively. They are treated as parts of the observed data sets as well as the accelerations. By using ICA, the similar shape to the input irregularity is obtained, and it is proven that the ICA is effective to monitor the railway system.
