Tuning the control penalty factor of a minimum variance adaptive controller

This paper presents a startup and tuning procedure for an adaptive control system based on a minimum variance controller, proposing a successive dual setting of the control penalty factor (ρ) which weights the influence of the control variance in a cost criterion function. The main idea consists of a differentiated approach of the transient startup regime (including the following steps: open-loop initialization of the parameter estimates, initial tuning of the controller parameters, closing the loop, synchronization between the controller and the plant, retuning of the control penalty factor), respectively of the stable steady-state regime reached by the control system, by tuning the controller using two distinct values of factor ρ. Firstly, in the control system startup phase, characterized by strong transient regimes which could destabilize the system, a high value of ρ factor is chosen. Such a value of ρ ensures a strong penalization of the control and, at the same time, smoothes the transient responses, leading the control system to a stable steady-state regime. However, a too high value of ρ factor is not proper for an efficient rejection of disturbances which perturb the plant during this stable state. Based on these considerations, a retuning of the controller is required through a significant decrease of the ρ factor, thus increasing the dynamic of control, which becomes able to provide an efficient rejection of the disturbances acting on the plant. The new proposed startup and tuning procedure was tested and validated for the case of an excitation control system of an induction generator connected to a power system. The novelty of the paper consists of this dual setting of the control penalty factor (ρ).