Railway anti-yaw suspension monitoring through a nonlinear constrained estimator

The anti-yaw suspension is a fundamental component of modern high-speed railway vehicles. Its application allows to reduce undesired vehicle hunting oscillations and increases the running stability. Typical calendar-based maintenance strategies are applied to repair or substitute parts of the anti-yaw suspension with a consequent increase in operating costs due to downtime. In this paper, a model-based strategy to monitor both the components of the anti-yaw suspension, consisting of spring and damper, is presented. A constrained unscented Kalman filter, coupled with a random walk model to estimate the wheel-rail contact interactions, is employed in order to apply constraints to the state variables allowing a more reliable parameters estimation. The chosen estimation technique allows obtaining a tool suitable for real-time applications to monitor the anti-yaw suspension health conditions taking into account the operative limits of its components, avoiding a priori knowledge of the track wear conditions.