Robust Zonotopic Set-Membership Approach for Model-Based Prognosis: Application on Linear Parameter-Varying Systems

Abstract

A robust set-membership Prognostics and Health Management (PHM) methodology is presented in this paper. The key advantages of the set-membership approach for states and parameters estimation are enhanced by employing zonotopes that are less conservative and computationally complex than other sets. The optimal tuning of the proposed observer is formulated using the Linear Matrix Inequality (LMI) approach. Moreover, the Joint Estimation of States and Parameters (JESP) leads to a non-linear representation of a monitored system that is transformed into a Linear Parameter-Varying (LPV) system by means of the non-linear embedding approach. The considered case study is based on a slowly degraded DC-DC converter. The aim of the proposed PHM approach is to forecast the Remaining Useful Life (RUL) on a system level. Additionally, the proposed RUL forecasting approach is independent of previous knowledge of the degradation behaviors being only dependent on the estimated zonotopic parameters. Finally, the obtained results demonstrate the efficiency of the proposed approach.