An efficient stochastic qualitative reasoning method for real time fault diagnosis of building air conditioning system

Presents an efficient qualitative reasoning method. During the diagnosis process how the qualitative reasoning reflects the real word is estimated by a parameter, called agreement rate. At every point of time, the agreement rate is calculated based on a series of behaviors derived over a period of time from that point. Because the computation of agreement rate for every point of time is necessary for diagnosis, this causes inefficiency of reasoning several times at each point. Therefore, we propose an efficient qualitative reasoning method to do the reasoning from the beginning to the end continuously. This method computes the agreement rate for each point successively while reasoning. In addition, this method incorporate a rapid behavior deriving technique in the continuous reasoning