Investigation on the influence of observation scheme of vehicle vibration on track irregularity estimation

Track irregularity is the primary excitation source of the running trains, impacting both operational safety and passenger comfort. Swift and accurate dynamic detection of track irregularities holds significant guidance for track maintenance and upkeep. Track irregularity can be estimated theoretically based on the Kalman filter method by taking the vibrational responses of the vehicle as observations. Owing to the complexity of the railway vehicle model and the limitation of the vibration sensors in quantity and accuracy, more investigations should be conducted on the practical sensor arrangement taking both accuracy and economy into account. In this study, a single vehicle model with 10 degrees of freedom was established, and the corresponding state space equation was deduced. The track irregularity was assumed to meet the random walk model so it can be estimated based on the extended Kalman filter. The reason of numerical drifting in the estimated track irregularity was discussed, and a method was proposed for eliminating the drifting phenomenon based on the ensemble empirical mode decomposition method. The accuracy of the present model and algorithm were validated through numerical examples. A total of 6 observation schemes were proposed considering the feasibility of sensor installation. Investigations were then made of the influence of different observation schemes on the estimated track irregularity. To validate the performance of the proposed method, numerical conditions with different variables were taken into consideration, including vehicle parameters, speed, and sensor noise. The calculated results show that the lack of the observation of the absolute displacements of the vehicle leads to drifting of the estimated responses of the vehicle. The observation schemes including three or more sensors generally give accurate estimation of the track irregularity. The observation of relative displacement within the vehicle has positive influence on the estimation of track irregularity of long wavelengths, but has negative influence on that of short wavelengths. Small deviation of actual vehicle parameters and speed from the theoretical ones only leads to estimation error of track irregularity in specific wavelengths. The signal-to-noise ratio of the measured vehicle response is positively correlated to the estimation accuracy in full wave band of the track irregularity. The findings in this study provide guidelines for arrangement schemes of on-board sensors for track irregularity estimation. © 2023, Central South University Press. All rights reserved.