This paper proposes a new method for the selection of input-output pairing in decentralized control structures for multivariable systems. This method proposes the input-output pairing problem as a multi-objective optimization problem (MOP). For each control structure and loop pairing analyzed, a different design concept is proposed and a MOP is stated. All MOPs share the same design objectives, and Pareto fronts associated with each design concept can be compared globally under a multi-objective (MO) approach. The design objectives were chosen for the MOP, as well as the designer's preferences, have an important role in selecting a certain loop pairing. The main contribution of the proposed approach is that it enables a systematic analysis of the conflicts between the objectives and the performance of a control system. The method enables selecting a certain input-output pairing and a suitable tuning of the controller directly using information that a designer can interpret. To show the application of the methodology, two loop pairing examples are presented, one of them for a two-input-output system (with four scenarios of analysis), and the other for a three-input-output system (with one scenario of analysis). Through the examples presented in this paper, it is evident how the designer can affect the loop pairing to be used, either by choosing the objectives or preferences.
- Multivariable control system
- Pareto front
- decentralized control structures
- input-output pairing
- multiobjective evolutionary optimization