Abstract
Common LMI solutions are used in most multiobjective control designs, however, common LMI solutions cause conservatism. In order to reduce this conservatism, this paper presents a new multiobjective gain scheduled control design method based on non-common LMI solutions for a descriptor form LPV system. In the proposed method, the selection of descriptor variables is devised, and the feedback gain is constant in appearance. As a result, parameter-dependent LMIs are able to be transferred to the finite number of LMIs without conservatism. The proposed design method is applied to an inverted pendulum, and its validity is shown through experiments as compared with the usual design method based on common LMI solutions.
