This thesis presents adaptive algorithms to identify and control systems with switching behavior and large uncertainties. First, adaptive identification algorithms for switched systems are derived. Hybrid observers are used to reconstruct partitions, parameter identifiers are extended form linear systems to identify the affine subsystems, and concurrent learning relaxes requirements on persistent excitation. Then, direct and indirect model reference adaptive control laws are proposed to make switched systems behave linearly in closed loop. Finally, a coverage-control algorithm for candidate-controller distribution in multiple models adaptive control is developed.
«
This thesis presents adaptive algorithms to identify and control systems with switching behavior and large uncertainties. First, adaptive identification algorithms for switched systems are derived. Hybrid observers are used to reconstruct partitions, parameter identifiers are extended form linear systems to identify the affine subsystems, and concurrent learning relaxes requirements on persistent excitation. Then, direct and indirect model reference adaptive control laws are proposed to make swi...
»