Model Invalidation for Switched Affine Systems with Applications to Fault and Anomaly DetectionF. Harirchi and N. Ozay This paper addresses the model (in)validation problem for switched systems with affine state space models. We recast the model invalidation problem as a mixed-integer linear program and solve it using state-of-the-art MILP solvers. In addition to traditional application of model invalidation to build trust in the models obtained from system identification, we illustrate how the proposed model invalidation approach can be used for anomaly and fault detection. Finally we show that knowing the fault model can reduce the complexity of online fault detection if the fault and system model satisfy certain conditions, which can be algorithmically and relatively efficiently verified offline. |