Decentralized semi adaptive control with switching and repeating offline identification

Abstract

This article presents some decentralized regulation approach, combining adaptive and switching. Offline identification is part of the method, which consists of several sub-steps, namely an initiation phase based on the identification of all sections when the desired values are changed, which is followed by repeating the identification when changing the weights for individual sections of one input and repeating this phase with a change of weights for all combinations of other inputs and then the weights are set to calculate the estimate of the system parameters according to the identification results. It includes PID controllers using internal model control method to compute the parameters and utilizing the least-squares method for identifying parameters. Parameters of system are estimated from previous call founded on every mixture of the set points, with some weight of their changes. The validity of the approach was confirmed on a system featuring 3 inputs and 3 outputs with interconnections implemented in the model's V-structure.