Using the μ-interaction measuring in the design and stability analysis of decentralized control structures

The results of using the μ-interaction measure as an analysis tool in systematically designing decentralized controllers for multivariable systems are given. The μ-interaction measure serves as the upper bound for the maximum allowable gains of a diagonal controller such that stability of the closed-loop diagonal system also implies stability of the closed loop full system. Several examples found in the literature are used to demonstrate this analysis technique. The diagonal controllers are designed using the internal model control method. An investigation has also been carried out to study the incorporation of model uncertainty into the μ-interaction measure analysis. Results indicate that, although the μ-interaction measure may render a somewhat conservative design, it does provide an effective means of assessing the stability of the full system when the controller design is based on the diagonalized plant system