On reducing compensator bandwidth of LQG/LTR control: an H∞- optimization approach

Due to lack of robustness, the Linear Quadratic Gaussian (LQG) approach to the design of control systems is enhanced by Loop Transfer Recovery (LTR) techniques. However, the robustness is recovered with the price paid to the unduly large or infinite bandwidth of the compensator. Such compensators are undesirable because plant output measurement noise is greatly amplified so as to degrade the system performance. A simple method is presented for reducing compensator bandwidth by H_∞- optimization of a stable function which combines the usual sensitivity function difference with a weighted closed-loop transfer function from noise to plant input. An example is given to show the improvement achievable by this new design method.