Application of particle filter technique to online prognostics for solenoid valve

Solenoid valves (SVs) are used as actuators in various applications, which are crucial parts in control system. Their failure may result in a system crash, so its health condition and reliability are related to the safety of an engineering system. In order to explore the basis of condition-based maintenance or replacement of SVs, it is necessary to develop a prognostic approach to predict its operation condition and remaining useful life (RUL). In this paper, a particle filter (PF) technique, an online tracking method, is proposed to make prognostics for SV. Moreover, the Brownian motion degradation model is proposed, and the distortion of dynamic driven current curve is assumed as the indicator of degradation state. To validate the proposed PF-based prognostics method, a degradation experiment is designed. The result shows that the predicted degradation state accurately reflects the real time filtered degradation state of SV and a good RUL prediction can be calculated by this method.