Plug-in module for controller reconfiguration based on latent variables and the Youla-Kucera parameterization

Abstract

This paper presents the design of a plug-in module to address the problem of controller reconfiguration in industrial processes. The proposal is based on a multi-controller switching philosophy, where the modification of an interpolation signal defines the combination of the control actions of each controller. The contribution is based on the integration of two methodologies. On the one hand, a multivariable feedback control design approach, using the concepts of control allocation and measurement combination. On the other hand, the mapping of a set of linear stabilizing controllers onto a multi-controller, based on the Q-parameter from the Youla-Kucera theory. In this context, the set of controllers can be designed independently. Moreover, the stability is guaranteed subject to an arbitrary switching between different stabilizing controllers. The procedure is evaluated by considering two relevant scenarios of control reconfiguration: 1- a complete modification of the input–output pairing, and 2- the replacement of a classical controller with a new advanced control strategy. Based on the computational simulation of two case studies from the literature, it is shown that the plug-in module carries out the reconfiguration of the control structure, improving the dynamic performance and ensuring the stability of the system. The design is based on the nominal controller, which is not modified during the reconfiguration process. In addition, it can be easily implemented online, connected to input–output terminals of the existing controller.