Nostra: Power System Condition Monitoring and Prognostics

Complex electro-mechanical systems will inevitably fail. These failures can be mitigated by modeling the system and predicting failures. We propose to build Nostra, a system that takes knowledge from domain experts as well as observed data and creates a model-based inference system for predicting failure in electronic systems. The Nostra system consists of three parts: a model, an inference engine, and an action selection component. The model, constructed using knowledge of key power system elements, like battery chemistry and charging algorithm, encodes both the nominal operation of the power system as well as common failure modes.