Diseño de sistemas de control basado en especificaciones de rendimiento y con consideraciones de robustez

This Project deals with the elaboration of a controller design framework being the main introduced feature to take into consideration the implications that arise as the result of the performance specifications. When a desired level of performance is specified for a certain system, it can not always be achieved. At least, not with a certain controller structure. This is why we believe it is necessary to know the implications on the minimum level of controller structure complexity and the minimum level of process information needed to achieve these performance levels. On the other hand, if the controller structure and complexity are previously know and fixed it would help to know the performance levels we can expect. This is why the framework is called performance driven. One the other hand, as the representation possibilities of a model and the actuation capabilities of a controller are limited, it is necessary to introduce robustness considerations into the design process. These considerations have to be included and the additional constraints they impose characterized. Considering also in what amount the performance is degraded with respect to the nominal or no uncertain case. The elaboration of such framework is based on: 1. Control architecture, and associated methods, where the performance implications and dependencies are explicitly shown. In addition, quantitative expressions for the minimum levels of information interchange among players and information processing capabilities needed to achieve a prescribed performance level. 2. Use the ideas and framework settled in 1) to the special case of polynomial systems and quadratic performance specifications. 3. Take advantage of the PID controller design approach based on min-max optimisation and using a polynomial approach (proposed by members of the research group) in order to give a practical viewpoint to the development results. The proposed methodologies are to be applied to test systems. (...)