LQG Vibration Control of Aircraft Panel Using Automated State Weighting Procedure
Optimal control theory has long been plagued by its inability to optimize the state and control weighting matrices specified in the LQR (Linear Quadratic Regulator) cost function. Although this control is optimal for a given set of user-defined state and control weighting matrices, the performance of the controller varies widely based on the selection of these weights. Engineers have been left with the task of choosing appropriate weighting sequences by iterating each weight until the controller performs to their satisfaction. This procedure gets increasingly more frustrating and time consuming as the size of the controller increases. The work in this paper outlines an effective strategy which reduces the engineer’s effort in finding these weights. It is shown that the introduction of an additional performance index along with the repeated perturbation of an initial guess quickly leads to a controller with excellent performance. Comparison of this automated method with a controller exhaustively designed using a standard selection method reveals a closed loop system response which has more energy reduction and more maximum amplitude reduction, all in a fraction of the time it takes to guess and check the weights. The controller will be applied to the model of an aircraft panel in an effort to reduce vibrations caused by wind. The goal is to achieve a reduction in cabin noise by controlling the vibration of such panels. The system identification procedure uses a modification of the SOCIT toolbox to achieve extremely accurate frequency domain system models. The model obtained using this method will then be used in the design and simulation of both the trial-and-error state weighted controller and the automated state weighted controller.