Relaxed Set-Membership Estimation and Cyber Attack Detection for LPV Systems Under Multiple Attacks via A Switching-Type Scheme Design

This article investigates the attack detection problem for linear parameter-varying (LPV) systems with unknown but bounded (UBB) noise, which are subject to multiple attacks. Traditionally, the control problems of LPV systems involve the introduction of the one-order free-weighting matrix (OOFWM) method, which is conservative and inevitably reduces the accuracy of estimation. In such cases, the SHOFWM method is proposed to address the conservatism problem. To be specific, the developed method exploits the parameter information inherent in the system and designs a more flexible free-weighting matrix. By further incorporating a switching mechanism, the SHOFWM method enables more degrees of freedom in calculating the constraint sets. This allows for relaxed conditions when determining the radius of constraint sets. Therefore, the constraint sets can be smaller, resulting in higher accuracy. Furthermore, an improved zonotope-based set-membership attack detection method is proposed. By utilizing the detection algorithm, it is possible to detect multiple attacks. Finally, the proposed methods are validated via two examples.