摘要：AbstractThis work proposes a real-time scheme to monitor the occurrence of faults and perform fault characterization. Faults, in this context, correspond to changes in the parameters of the system being monitored. The method relies on the concept of Analytical Redundancy Relation (ARR), which can be defined as the evaluation of the mathematical constraints of the physical model of the system given the real, noisy measurements. The algorithm consists of two modules: a detection strategy that relies on the regular application of an ARR-based hypothesis test in discrete time-steps; and an optimization procedure to estimate the changes undergone after a fault. By selecting a set of feasible solutions from the output of the optimization algorithm, the method also sheds some light on the uncertainty associated to the estimated quantities. The methodology is tested on simulated data of lithium-ion batteries in unmanned aerial vehicles.
