Stability Analysis of Three Types of Uncertain Neural Networks with Time Delays
|Course||Operational Research and Cybernetics|
|Keywords||Neural Network Stability Analysis Linear matrix inequality Delay Pulse|
The neural network is a highly integrated interdisciplinary , it is able to simulate human intelligent behavior . In recent years, the neural network has become one of the \Stability of the system to normal operation of the premise , and robustness is the key to survival in the case of abnormal and dangerous system , so the study of neural network system stability and robustness of great significance . Papers using the Lyapunov stability theory and linear matrix inequality technique to study the stability of the three types of uncertain time-delay neural network problems , ensure the stability of the system . Reduced as compared with the existing results conservative . First, Uncertain Stochastic Neural Networks Robust Stability in Mean Square . A class given in the form of linear matrix inequalities , delay-dependent global robust mean square exponential stability criterion is obtained by constructing a new Lyapunov function . The numerical example illustrates the effectiveness and superiority of the criteria given . Secondly , we study a class of a Markov chain with time-varying delay pulse random Cohen - Grossberg neural network model , the application of some inequality techniques and stability theory , model robust mean square gradually theorem nearly stability . The stability theorem has better applicability . Finally, simulation examples demonstrate the validity of the theorem . Again, for a class of discrete random neural network model with discrete and distributed delays , by constructing a new Lyapunov function gives the the model asymptotic stability theorem . Theorem, given in the form of linear matrix inequalities easily solved using Matlab LMI toolbox . Finally, simulation examples demonstrate the validity of the theorem .