Approximate Q-learning approach for Health Aware Control Design

Health-aware control (HAC) has emerged as one of the domains where the control is formulated based upon the failure prognostics of system/component or the Remaining Useful Life (RUL) predictions of critical components. Usually, the degradation models are available for components under degradation, but accurate dynamics of the global system is not available. Moreover, mathematical dynamic (transition) models of RUL are rarely available, making the incorporation of RUL information in the control paradigm a difficult task. This paper proposes use of Approximate Reinforcement Learning based algorithms for control formulation. In particular, the approximate Q-learning based algorithm is developed to obtain optimal control law that manages the degradation speed in order to satisfy desired RUL requirements. The proposed method is studied using simulation of a DC motor and degradation ofbearing mounted on shaft.