Subsystem Reduction for Qualitative Simulation

Qualitative simulation (QS) is an area of artificial intelligence that represent continuos and discrete aspects like space, time and quantity with little information, and makes inferences using symbolic data to represent physical quantities. Tradicionally, QS uses a global statebased representation to represent the behavior of the system. To qualitative simulate a system, some initial values are normally given, along with qualitative differential equations (QDEs). With this information, a qualitative simulation algorithm evaluates all possible combinations of qualitative values and filters out inconsistent states considering qualitatives constraints. This is a combinational process which normally requires exponential time. In this paper, a new algorithm is described which simulates individual components independently and joins their behavioral graphs together until a global behavioral graph is obtained. This algorithm achieves substantial reductions in time and it is polynomial with respect to the number of components. It is shown how with this algorithm is possible to simulate industrial plants of hundreds of variables within a few minutes.