An analysis of stability of uncertain neural networks with multiple time delays

This paper deals with the problem of robust stability of neural networks with multiple time delays with the class of unbounded and nondecreasing activation functions. By constructing a suitable Lyapunov functional and applying the homeomorphism mapping theorem, we derive new delay-independent sufficient conditions that establish the existence, uniqueness and global asymptotic stability of the equilibrium point for the delayed neural networks under norm-bounded uncertain network parameters. The conditions obtained for robust stability are expressed in terms of network parameters only, therefore they can be easily checked. An advantage of the proposed results is that they consider the number of the neurons in the stability conditions. We also give some numerical examples with comparative results to demonstrate the applicability of our stability conditions. These comparative examples will also show the advantages of the obtained results over the corresponding robust stability results derived in the previous literature.