Enhancing Fault Diagnosis through Robust Model Reference Adaptive Control: A Set-Based Approach

A method for designing inputs in active fault diagnosis combined with a robust model reference adaptive controller is proposed in this paper. The approach ensures that the output of a plant with unknown parameters converges to a reference model during the diagnosis process. The paper proves that Lyapunov stability can be guaranteed, given a condition on the parameters of the robust adaptive law. Additionally, a novel set-membership filter, employing constrained zonotopes, is introduced. This filter is computationally efficient and does not require outer approximations. The active fault diagnosis is accomplished by generating reference inputs online, allowing the output sets to separate for different plant models. A demonstration of the proposed method on a numerical example is provided in the final.