Compatible Control for Multi-objective System Based on Linear Matrix Inequality
Qingsong Hu, Lihong Xu, Zhiqiang Zou · 2006
In a real multi-objective control plant, all the objects have certain acceptable range. Taking linear uncertain system as model, by adopting linear matrix inequality (LMI) optimization method, we research the compatibility of different objectives in multi-objective control problem. A two-layer controller design method is proposed, i.e. the first layer obtains the compatible interval objectives which meets the customer's demand best by compatible optimization and the second layer designs the controller using traditional methods based on the result of first layer. The new method overcomes the drawback of traditional methods which design the controller taking the "accuracy point" as objectives and are difficult to be adopted in practical application. This new design approach fills the gap between theory and application to a certain degree