Model-based multi-component adaptive prognosis for hybrid dynamical systems

Abstract A bond graph model-based prognosis method for multiple components with unknown degradation patterns in a hybrid dynamical system is proposed. The traditional approach for remaining useful life prediction with single degradation model is inappropriate for hybrid systems where the dynamics changes according to operating mode. Therefore, multiple degradation models are suggested and these are adapted with new information of the degradation states of the monitored system. Sensitivity-based dynamic signature matrix is utilized for degradation hypothesis generation which provides the deviation directions of fewer hypothesized degradation parameters and thereby accelerates parameter and degradation trend estimation. The results are supported by experiments.