摘要：AbstractOffline parameter identification as well as online estimation of state variables are common tasks for a large variety of control applications. These tasks become especially challenging if parameters are not known with absolute accuracy. Then, interval representations can be used to describe worst-case bounds of uncertain parameters and to compute guaranteed enclosures of the trajectories of all state variables. In this paper, an interval-based parameter identification procedure is presented, extended to an observer approach, and validated experimentally for a finite volume representation of a distributed heating system. For the implementation of this procedure, the structural property of cooperativity is exploited which allows for constructing two independent sets of state equations describing guaranteed lower and upper bounds of all system states.
关键词：KeywordsDistributed-parameter systemsearly lumping techniquesfinite volume modelinginterval observersparameter identificationcooperativitypositive systems
