Automatic parameter tuning of stochastic qualitative model of building air conditioning systems

Abstract

We have proposed the stochastic qualitative simulation which can derive approximate behavior from a simple qualitative model of a target. In this method, the model must be constructed with numerous stochastic parameters. The parameter tuning process is the most difficult element of model construction. This paper outlines an automatic parameter turning by means of the steepest ascent based method. This method was used in order to generate a model of a real air conditioning system in a building.