On qualitative identification of linear dynamical systems

The paper proposes a method for identifying qualitative models of linear dynamical systems. First, the state variables of a given system are measured qualitatively, i.e. it is only known in which region of the state space the state vector resides at a given instant of time. In the second step, the proposed algorithm gives bounds for the system matrix of the linear state space model by methods of interval arithmetic. In the last step, the qualitative model in form of a nondeterministic automaton is determined from the identified interval system matrix.<
>