Vibration-Based Estimation of Wheel-Rail Contact Forces Using an Extended Kalman Filter for Wheel Condition Monitoring

Purpose This study aims to develop a vibration-based inverse approach for estimating wheel-rail contact forces from axle acceleration measurements, addressing challenges associated with nonlinear contact behaviour, variable operating speeds, and stochastic track irregularities. Methods A reduced-order vertical vehicle-track interaction model is developed to represent the dynamics of the wheelset, rail, and supporting track structure. An Extended Kalman Filter (EKF) reconstructs contact forces from measured axle acceleration signals. Reference responses are generated using a detailed 3D multi-rigid-body vehicle model implemented in the VAMPIRE simulation software, incorporating the standard S1002 wheel profile and UIC60 rail profile. The estimation performance is evaluated across different operating speeds and wheel-flat severities. Results The proposed method reasonably reconstructs wheel-rail contact forces within the dominant vibration frequency range relevant to vehicle-track interactions. The estimated forces capture variations associated with wheel-flat severity and vehicle speed, demonstrating the potential of the approach for wheel condition monitoring. Conclusions The proposed vibration-based inverse method provides a promising approach for estimating wheel-rail contact forces using measurable acceleration signals. Its potential integration into Digital Twin frameworks could support predictive maintenance, improved defect detection, and enhanced railway safety and reliability.