Minimax-LQG control of a flexible plate using frequency domain subspace identified models

In this paper identification and optimal robust Minimax-LQG control design are studied for active vibration control of a flexible plate. Two frequency domain subspace methods including maximum likelihood technique are used to identify the model of the plate. Since the identified models are unstable, an iterative algorithm is applied to stabilize them. The obtained models have a good fitness up to the frequency 220 Hz and this frequency range contains three modes of the flexible plate. These first three modes are selected for control and the rest are chosen as uncertainty. The Chebychev and Yule-Walker filters are designed to model the weighting function of uncertainties for design of Minimax-LQG controller. These weights have a great effect on robust stability and performance of the control system. Simulation results are presented to show the effectiveness of designed controllers for all stabilized identified models and uncertainty weights. The results confirm that the obtained indices for identified models are good measures to predict the performance of the designed controllers based on them.