Multiple models for adaptive control to improve the performance of minimum variance regulators

The design of a minimum variance regulator for systems operating in dynamical uncertain environments is considered. For minimum variance self-tuning control of an unknown linear time-invariant system, a stable control strategy is developed that improves the transient response by using multiple models to describe the different environments and switching between the relative controllers. The controller is determined at every time instant by the model which best approximates the plant. The performance of the controller is evaluated by a set of simulation tests.