Interval Estimation for Switched Neural Networks by Zonotopes

ABSTRACT This article investigates the zonotopic state estimation for a class of discrete‐time switched neural networks (SNNs) with mode‐dependent average dwell time (MDADT)switching. First, by building appropriate mode‐dependent Lyapunov functions, some sufficient conditions are established to guarantee the stability and performance for the considered system. Utilizing the formulated criteria, we propose a codesign scheme for switching signals and the mode‐dependent nonlinear observer. In addition, a time‐varying state zonotope is constructed, and with this as the basis, estimated bounds are calculated accordingly. Eventually, an illustrative example is utilized to demonstrate the efficacy and superiority of the results derived in this article.