Interval Identification - a Modelling and Design Technique for Dynamic Systems

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 first 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 identification 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 ofa chemical reaction system and safety assessment of an exothermic chemical process. Theexamples are based on real industrial and laboratory data.