State Estimation And Stability Analysis For Neural Networks With Markov Jump And Distributed Delays

In this thesis, the stability analysis and state estimation for neural networks with Markovjump and delays were researched. Some new delay-dependent asymptotic mean-square stabilitycriteria for neural networks with delays have been proposed. Then, the state estimators weredesigned. The results were given in the forms of LMIs, which can be easy to solve by the LMITools in MATLAB. Some numerical examples have been provided to illustrate the effectivenessof the aforementioned results. The main content of this thesis can be presented as three parts:In the first part, the stability analysis and state estimation for neural networks with Markovjump and distributed delays were researched. By choosing proper Lyapunov-Krasovskii func-tional, the asymptotic mean-square stability condition of the neural networks was given. Then,astate estimator was designed. The results were given in the forms of LMIs. Some numericalexamples have been provided to illustrate the effectiveness of results.In the second part, the neural networks with Markov jump and distributed delays have beenresearched. The parameters uncertain were also considered. By choosing proper Lyapunov-Krasovskii functional, the asymptotic mean-square stability condition of the neural networkswas given. The conservation of result in this part was less than the former. Then,a state estimatorwas designed.The results were given in the forms of LMIs. Some numerical examples have beenprovided to illustrate the effectiveness of results.In the third part, the neural networks with Markov jump and delays have been researched.The transition probability considered in this part was partly unknown. Based on the character-istic that the sum of transition probabilities was zero, a new asymptotic mean-square stabilitycriterion was proposed by appropriate Lyapunov-Krasovskii functional.Then,a state estimatorwas designed. The results were given in the forms of LMIs. Some numerical examples havebeen provided to illustrate the effectiveness of results.