Application of Kalman filter to Model-based Prognostics for Solenoid Valve

Solenoid valves (SVs) are electromechanical components, which are used as actuators in various application environments and play crucial roles in control systems, and their breakdown may result in a system crash. Therefore, this paper explores a Kalman filter (KF)-based method to predict the remaining useful life (RUL) of SVs, so that the SVs can be replaced or maintained before their failure bringing a catastrophic consequence for engineering system. In this paper, a degradation signal is extracted from the driven current, which can be monitored conveniently with a non-contact current sensor. Based on an empirical linear degradation model, the KF is adopted to track the degradation state and the degradation rate and to capture the uncertainties. The Monte Carlo sampling and kernel density estimation are used to propagate the uncertainties and estimate the probability distribution of RUL, respectively. To verify our methods, a degradation experiment is designed. The experiment results show that the degradation signal extracted from the driven current can indeed reflect the degradation state of SVs. By comparing the proposed method with other state of the arts prognostic approaches, it shows that the proposed KF method preforms better and has a higher prediction accuracy than other methods.