凸最適化によるデータに基づく制御器設計の展開

抄録

Recently, many researches extracting an essense from a huge number of data receive much attention in the various field of engineering. In the field of control engineering, the controller sysntheses from time-domain and/or frequency-domain data are energetically studied based on numerical optimization techniques. This paper presents a recent development of data-based controller synthesis by convex optimization, and shows that a special case of nonconvex optimization problems, which is often encounterd in the control field, is reformulated as a convex optimization problem by introducing the Concave-Convex Procedure or the Iterated LMI constraints. As an application, the H_∞ loop shaping design procedure reformulated using frequency response data of a controlled plant is presented.