The goal of this project is to develop an efficient and reliable methodology for the simulation of uncertain and qualitative models of dynamical systems.
This project proposes a methodology of simulation for a more general notion of dynamical system where the domain of the state variables is extended from the set of real numbers to the domain of uncertainty distribution (bayesian and non) and the behavior of the system may be modelled not only by differential equations but also by qualitative relations (e.g. inferential rules and qualitative networks). The innovative aspect of the research consists in the applications of uncertainty techniques and representations, developed in artificial intelligence, to the modelling and simulation of dynamical system, treated in system theory. The expected result is a set of computational methods to simulate and forecast the behaviour of a system whose model may be described both in quantitative (algebraic and differential mathematical equations) and in qualitative terms (inferential and network relations) and where probabilistic and non probabilistic formalisms may be used to represent the uncertainty.