Switching Between Controllers Using Replicator Dynamics

We propose replicator dynamics to design an alternative algorithm for switching control. We use a feedback control strategy with a finite set of commuting controllers, whose control action takes a linear and time-invariant (LTI) process output to a desired set point. Each controller represents a pure strategy of the evolutionary game, in which case the assigned resource is the time allowed for each controller to operate the process. Therefore, from an evolutionary game theoretical perspective, we give a higher payoff to those controllers with the best performances to control the process according to a specific fitness function. Our proposed control strategy is compared in simulation to an existing switching system strategy, and it preserves the stability conditions that have been developed for such systems.