Based On The Unified Control Theory Of Dissipative Structures And The Lmi Framework

At present, with the development of cybernetics, its embranchments are more and more, its models are more and more diverse, and its algorithms also change more and more complicatedly. Concerning this phenomenon, based on energy dissipative property and LMI convex technique, the unified problem of control theory to uncertain system is investigated in the dissertation. Also, the application problem of its analysis and design method is discussed. Some research work on the problems mentioned above is listed, as follows:1. Dissipative property and dissipative inequality in linear and nonlinear system are studied. The relation between different robust performances, control parameters and dissipative inequality is presented and is described in a unified maths formula. Namely, the main equation under unified framework via various control problems is built.2. The theory and methods based on LMI are studied, and are combined with calculation technique of numerical value, thus a unified linear algebra method is put forword. As a result, various control problems can be reduced to the problem solving three LMIs. Namely, the system algorithm under unified framework via various control problems is built.3. According to the theory and method under unified framework, robust control problem and robust filtering problem are studied. First, robust controllers design with mixed performance requirement is researched by giving output-feedback synthesis with H2 , H and regional pole placement constraints. Second, reduced-order robust filters design for continue-time/discrete-time systems is researched. The filters obtained are a set of permitted ones and have optimality and unbiased estimation.4. Considering ordinate navigation of certain torpedo system as applied object, output-feedback controllers with H2 , H and regional pole placement constraints and reduced-order robust filters are designed. Furthermore, simulation results show efficiency and feasibility of the theory and algorithm researched in this paper.