International Workshop on Qualitative Reasoning , June 6 - 9 1999 , Loch Awe , Scotland 1 Interval Identi cation - a Modelling and Design Technique forDynamic

A new technique is presented to identify intervals for parameters and initial conditions for nonlinear dynamic systems based on an imprecise mathematical model and measurements of system variables. This technique employs a fuzz interval qualitative simulator for interval dynamical models and a qualitative description of measured signals , the episodes. These are based on the sign of the variable and of its rst and second derivative. Episodes describe in a very simple and intuitive form the dynamic behaviour of the measured variable. By combining this qualitative with numerical information, a structure and parameter identiication can be done in an very intuitive and explainable way. This technique is used within a tool, TAM-C, to model and assess chemical processes involving exothermic chemical reactions. Two applications of this technique are given: modelling of a chemical reaction system and safety assessment of an exothermic chemical process. The examples are based on real industrial and laboratory data.