Objective Current wheel-rail force acquisition equipment faces challenges such as difficult data collection and high operational costs. Therefore, it is necessary to conduct research on wheel-rail force inversion algorithms.

Method Based on bogie vibration signals, a time-domain identification method for the left/right vertical wheel-rail forces and the lateral wheel-axle force is proposed. Frequency-domain integration is performed on the axle box and frame accelerations to obtain the velocity and displacement responses of the primary springs. Simultaneously, combined with the wheelset motion equations, the wheel-rail force inversion algorithm is derived. A vehicle-track coupled dynamics model is used to obtain the forward-calculated wheel-rail forces, axle box, and frame accelerations. The acceleration signals are used as input for the algorithm to compute the inverted wheel-rail forces, which are then compared and analyzed against the forward-calculated forces.

Result & Conclusion  The wheel-rail forces obtained by the proposed inversion algorithm and the results from dynamics model calculation show that the correlation coefficients of vertical and lateral forces are 0.74 and 0.89 respectively for straight tracks; 0.64 and 0.66 for under-superelevation curved tracks; and 0.60 and 0.65 for over-superelevation curved tracks. All three conditions indicate a strong correlation, demonstrating that the proposed wheel-rail force inversion algorithm achieves high identification accuracy for wheel-rail forces. At train speeds of 70 km/h, 80 km/h, and 90 km/h, the errors in the derailment coefficient and wheel load reduction rate indicators from the forward model and the inversion model are all less than 15%, indicating that the proposed algorithm maintains good identification accuracy at all different speed levels.