Identification of discrete event models for continuous-variable systems

The paper proposes a method for generating a discrete event model for a continuous-variable plant from experimental data. Measured sequences of qualitative values of the state variables yield a linear interval matrix equation which can be used to find bounds for the parameters of a linear state space model. The qualitative model in a form of nondeterministic automaton is then constructed from the identified interval system. It can be ensured that the automaton generates all possible qualitative trajectories of the given system. The method is illustrated by an application.