Extended dissipative estimator design for uncertain switched delayed neural networks via a novel triple integral inequality

Abstract

This paper addresses the problem of extended dissipative estimator design for uncertain switched neural networks (SNNs) with mixed time-varying delays and general activation functions. Firstly, for dealing with triple integral term, a new integral inequality is derived. Secondly, based on the theory of convex combination, we propose a novel flexible delay division method and corresponding modified Lyapunov–Krasovskii functional (LKF) is established. Thirdly, a switching estimator design approach is contributed, which ensures that the resulting augmented system is extended dissipative. Combining the extended reciprocally convex technique with Wirtinger-based integral inequality, improved delay-dependent exponential stability criterion is obtained. Finally, a example with two cases is provided to illustrate the feasibility and effectiveness of the developed theoretical results.