Improving train position accuracy in case of wheel jamming faults
Résumé
This paper is about position estimation and fault detection under jamming fault for rail transportation system. The motivation is to improve the accuracy of the position estimator in real-world conditions, considering modeling errors, perturbations and expected faults. The approach is based on an unknown input observer used to estimate perturbations and detect jamming faults for a class of polytypic linear parameter-varying (LPV) system. Moreover, the mixed H∞/H_ approach is applied to help in the robustness of the fault detection via a residual. Offline experiments results are provided for illustration.