Shaping-Filter Augmented Constrained EKF for Equivalent Conicity Estimation of Railway Vehicles

Equivalent conicity is a critical indicator of railway vehicle stability and safety. However, facility-based offline measurements fail to capture the progressive degradation of wheel profiles between maintenance intervals, motivating a transition toward realtime online monitoring. Conventional approaches often assume known track inputs or fixed creep coefficients, which are practically infeasible or cause steady-state bias due to model mismatch. To address these issues, this paper proposes a stochastic constrained extended Kalman filter augmented by a shaping filter to model track irregularities as colored stochastic processes. By integrating this augmented state modeling with Fisher information-based constraint scaling, the proposed filter robustly identifies key contact parameters without explicit track inputs. Crucially, spectral analysis reveals that parameter observability is primarily driven by the vehicle’s natural resonance (4–7 Hz) rather than unstable hunting modes, ensuring reliable estimation during stable operation. Ultimately, this proposed real-time condition monitoring framework facilitates a reliable transition from periodic maintenance to condition-based maintenance, enhancing operational safety and efficiency.