Proposal of FCbT by Considering Closed-Loop Stability at Each Parameter Update

Abstract

We propose a controller tuning method that considers closed-loop stability after controller tuning without a mathematical plant model. We propose Fictitious Correlation-based Tuning (FCbT) that obtains reasonable controller parameters using a couple of input/output data. However, closed-loop stability after controller tuning is not guaranteed in this approach. In this study, we impose the stability constraint to the parameter update in order to maintain closed-loop stability at each parameter update. Moreover, by introducing particle swarm optimization (PSO) instead of the Gauss-Newton method to solve the constrained nonlinear optimization problem, the initial value dependence is considerably reduced. The effectiveness of the proposed method is confirmed by a numerical example and experimental results.