MULTI-CRITERION APPROACHES TO OPTIMIZATION OF LINEAR REGULATORS
Additional Document Info
American Institute of Aeronautics and Astronautics, Inc., 1986. All rights reserved. The problem of designing structural control systems is addressed, considering multiple design objectives state error energy, control energy and stability robustness. The relationships among the selected design objectives is investigated by solving multiple objective optimization problems, which require parameterizations of feedback gain matrices directly or indirectly. Considered are three types of state feedback design algorithms including a new one developed in this study: 1) generalized LQR design [1,2], 2) modified LQR design [new] based on Lyapunov equation and 3) feedback design using Sylvesters equation [5,6]. The results of these design schemes are compared by generating trade-off surfaces of the multiple performance objectives. A nonlinear programming method based upon minimum norm correction strategy is utilized in conjunction with a homotopy technique. The nonlinear programming algorithm is found to be very reliable. The paper offers a new approach to multi-dimensional tuning of each feedback design.