Design Of State Estimator For Neural Networks Based On The Sampled Data

In this paper, we mainly analyze the design of sampled-data state estimator for neutral-type neural networks and competitive neural networks.The full text is divided into the following chapters:In Chapter1, we mainly introduce some backgrounds about the researches and development of neural networks, and there are some studies about state estimation and sampled-data state estimation of the neural networks.In Chapter2, the state estimators are designed for delayed neural networks of neutral-type based on sampled-data. In this chapter, by using a delayed-input approach, the sampling period is converted equivalently into a bounded time-varying delay. Based on a suitable Lyapunov functional, a sufficient condition for the existence of state estimator is derived in terms of linear matrix inequalities (LMIs). Then, through certain matrix transformation techniques, we construct the second state estimator by utilizing partial information of the neural networks based on the available sampled measurement. This estimator is established using only a part of the information network model, and more subtle than the first method, the calculation process is easier, thus saving computational cost.In Chapter3, the sampled-data state estimator is designed for competitive neural networks with both leakage delays and time-varying delays. Based on a suitable Lyapunov functional and the free-weighting matrix method, a LMI-condition is derived to guarantee the existence of the sampled-data state estimator.After the results of each chapter, the numerical simulation examples are given to illustrate the effectiveness and the feasibility of the theoretical results obtained.