Passivity Analysis for Discrete-Time Neural Networks with Mixed Time-Delays and Randomly Occurring Quantization Effects.

Jie Zhang,Lifeng Ma,Yurong Liu
DOI: https://doi.org/10.1016/j.neucom.2016.08.020
IF: 6
2016-01-01
Neurocomputing
Abstract:This paper investigates the passivity analysis problem for a class of discrete-time neural networks subject to the mixed time-delays and randomly occurring quantization effects. Both the time-varying discrete delays and the infinite distributed time-delays are considered. The phenomenon of randomly occurring logarithmic quantization is taken into consideration, which is described by a random sequence obeying the Bernoulli distribution. Sufficient conditions are established, guaranteeing the globally asymptotical stability in the mean square and the strict (Q,S,R)−γ-dissipative property of the considered neural networks. The main results are proposed by virtue of the linear matrix inequality approach that can be easily solved by certain convex optimization algorithms. The obtained methodology is capable of being adopted in the passivity analysis with little modifications. A numerical example is provided to verify the correctness and effectiveness of the exploited methodology.
What problem does this paper attempt to address?