Unfalsified adaptive control: A new controller implementation and some remarks

The concept of unfalsified adaptive control using multiple controllers and switching ideas has been developed and extensively investigated over the past decade [1]-[16]. In this literature of unfalsified adaptive control, it is required that the controller set only contains linear, bi-proper minimum-phase controllers. In this note, we propose a controller implementation scheme which overcomes this restrictive assumption on the controller set though retaining linearity. Furthermore, we advocate caution in some circumstances where the unfalsified adaptive control algorithm actually connects up a destabilizing controller in the closed-loop for a long period of time, and explain how this phenomenon can arise.