Design Of Robust Filter For Uncertain Stochastic System

The filtering problem is an important part of control theory, and it has been one of the problems receiving much concern in the circle of control theory and project practical application all the time. Filtering theory is widely applied in the aerospace, marine, industrial process control. The state estimation methods are widely used in the well-known Kalman filtering theory and Luenberger observation theory, which require the model must be accurate. In some industry applications, however, when the system is subject to parameter uncertainties, the accurate system model is very difficult to be obtained. To overcome this difficulty, robust filtering approaches are proposed. The random factors exist objectively in the actual process, if we do not consider the impact on inherent random factor of the system, dynamic performance of systems will deteriorate and not meet the request of expecting. So it has important theory and engineering significance to study robust filtering problem of stochastic uncertain delay systems.Mainly based on Lyapunov stability theory, the paper adopt linear matrix inequality technology to study robust filtering problem of uncertain system described with the state space. Aiming at stochastic uncertain delay systems, we design H_∞and L₂- L_∞filter separately; H_∞filter for neutral stochastic uncertain delay systems; and the application of stochastic uncertain delay systems in the wind power generation systems.this article presents the H _∞and L₂- L_∞performance criterion for stochastic uncertain delay systems, deduces sufficient conditions of the existence of H _∞and L₂- L_∞filter, and expresses them into a form of the linear matrix inequality (LMI),then utilizes schur complement lemma transformed them into LMIs which can be solved. We utilize standard digital software to get filter parameters.The filter designed by this method can ensure the norm of transfer function is minimum from disturbance to error. Therefore the impact of disturbance on system error will be reduced, and the robustness of a system will be enhanced.