Control refinement for DAE systems: A behavioral approach via simulation relations

14 Mar 2017  ·  Chen Fei ·

The controller design of the so-called "difference algebraic equation" (DAE) systems that are frequently shown in industrial processes, tend to be challenging because of the combination of algebraic equations and high state dimensions. In this paper, we tackle this problem by developing control refinement approaches for DAE systems via the notions of (bi)simulation relations and approximate simulation relations from computer science. The quantified refinement accuracy is achieved by defining observation metrics over a general system framework named transition systems. We employ the behavioral theory to tackle dynamical systems and control problems in a more general framework. Due to the difficulty in dealing with a DAE system directly, we derive another system, which is behaviorally equivalent to the related DAE system and in standard state space form, to provide ease for further control refinement. Consequently, well-developed model reduction approaches can be applied to obtain an abstract simplified system, which can be rewritten into a DAE system again. Based on the (bi)simulation relations, approximate simulation relations and the initialization conditions, we show that for any given well-posed controller of the abstract model, we can always refine it to a controller for the concrete model such that the two systems have the same controlled output behavior or the distance between their output behavior is bounded.

PDF Abstract
No code implementations yet. Submit your code now

Categories


Systems and Control

Datasets


  Add Datasets introduced or used in this paper