Stability Analysis Of Neural Networks With Time - Dependent Dependence

Neural networks is a hot topic in recent years, it has been widely used in a number of research fields, such as signal processing, pattern recognition, associative memory state estimation, and so on, The stability analysis is the most important in learning the neural networks, especially in the field of power system.The main work of this paper is separately get the delays global exponential stability conditions for the neural networks the with discrete delay, discrete and distributed delays in the third and fourth parts.And discuss the state estimation of the neural networks with discrete and distributed delays in the last part of this paper.Firstly, we discuss the Stability of the Delay-dependent neural networks with discrete delays. Firstly, I make a simple introduction and system conversion, the equilibrium point of this system is to be transferred to the origin of coordinates; then by constructing an appropriate dimension function and the delay interval is divided.Features of this article is adding triple integral, and use convex combination, Jessen inequality and freedom matrix and other techniques to deal with the derivative function of the integral term, and eventually give the the new determination exponential stability conditions of the Delay-dependent neural networks System in the form of LMI, And test the feasibility of conclusions by experimental data.Second, we added the distributed delays. And the focus work of this section is to find global Global exponential stability conditions of the neural networks system with discrete and discrete delays. and we give a different Lyapunov function, the delay interval to be divided, discuss global exponential stability of delay-dependent neural network models at each interval, the new globally exponentially stable conditions are finally get in LMI, and the data resulting simulation conditions than the existing literature has less conservative.The third part of this article discuss the state estimation of the neural network systems with mixed delays by adding the output variables. because the delay interval we talked have upper and lower bounds of uncertainty, so the functional contains we constructed contain the bounds of the delay interval. Delays due to segmentation method used herein to discuss the stability of the problem, so we added the entry point to the function, the dividing point satisfy the bounds of the delay function, then useintegral inequality. Giving the exponential stability conditions in LMI. and data simulation.