Toward Practical Adaptive Control

Systems containing both real parametric uncertainty and weighted-ball-in-H^¿ uncertainty are addressed. The concept of robustness given perfect parameter information is reviewed. Robust parameter estimation laws designed for maximal convergence in the presence of non-parametric ball-in-H^¿ uncertainty are reviewed. It is shown that if the control laws would produce a robust design given perfect parameter information, and if the robust estimation laws are used in lieu of knowledge of the parameters robust bounded-L²-input bounded-L²-output (L²-BIBO) stability automatically results. Shortcomings of BIBO stability are discussed.