Robust Control Methods

This is the final report for research supported under AFOSR Grant F49620-95-1-0095 during the period December 15, 1994 through August 31, 1998. The research focused broadening class of solvable robust control problems and on developing a firm information theoretic foundation for incorporating the real-time effects of evolving experimental data in adaptive robust control system designs. Robust control concerns the problem of engineering control systems capable of robustly maintaining performance to within prescribed tolerances in the face of large-but-bounded modeling uncertainties and nonlinearities. Significant advances were achieved in developing Bilinear Matrix Inequality SMI robust control design methods. The BMI significantly expands the class controller design constraints that can be accommodated to include reduced order control, decentralized control, multi-model control, gain-scheduling, mixed H2H infinity control and so forth. In a separate development, a theory of unfalsified control has emerged as a precise tool for characterization and optimal utilization of the evolving information flows in adaptive control processes. This theory has also been demonstrated to lead to faster, more reliable adaptive control designs. The results are expected to be useful in advanced aerospace control applications where robust performance is prerequisite.