Relaxed Co-Design of Attack Detection and Set-Membership Estimation for T-S Fuzzy Systems Subject to Malicious Attacks

• Published in: IEEE transactions on fuzzy systems Vol. 32; no. 5; pp. 2663 - 2676
• Main Authors: Du, Mengni, Xie, Xiangpeng, Wang, Hui, Xia, Jianwei, Chadli, Mohammed
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.05.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Co-design; Detection algorithms; Fuzzy systems; Heuristic algorithms; Homogeneous polynomial; malicious attacks detection; Matrix methods; multimode mechanism; Noise measurement; Polynomials; Security; set-membership theory; State estimation; Switches; Switching; Weighting
• ISSN: 1063-6706; 1941-0034
• DOI: 10.1109/TFUZZ.2024.3357135

This article is concerned with the attack detection problem of Takagi-Sugeno (T-S fuzzy) systems with unknown but bounded (UBB) noise subject to malicious attacks via the set-membership estimation and a switching multimode high-order free-weighting matrix (SMHFM) method. Initially, the SMHFM is developed to address the so-called conservatism problem caused by the one-order free-weighting matrix method. To solve this problem, the homogeneous polynomial technique is employed to introduce groups of free-weighting matrices for different switching modes. As a result, the SMHFM is synchronized with the working modes and possesses a high-order feature. In addition, a switching multimode mechanism is implemented to enable the SMHFM to exhibit multiple modes. Time-variant balanced matrices are introduced for different switching modes to adjust the relevant matrix terms. This adjustment allows for obtaining more relaxed conditions, leading to smaller constraint sets and higher accuracy in state estimation. Furthermore, a zonotope-based set-membership attack detection algorithm is introduced for T-S fuzzy systems, which is capable of detecting various types of attacks. By utilizing the proposed SMHFM method for attack detection, the level of conservatism in state estimation can be reduced. Finally, two simulation examples are given and some comparisons are made to validate the effectiveness of the proposed methods.