GAUSSIAN MIXTURE DENSITIES FOR CLASSIFICATION OF NUCLEAR-POWER-PLANT DATA

In this paper we are concerned with the application oi learning algori thms to the classification of reactor states in nuclear plants. Two as pects must be considered: (1) some types of events (e.g., abnormal or rare) will not appear in the data set, but the system should be able t o detect them, (2) not only classification of signals but also their i nterpretation are important for nuclear plant monitoring. We address b oth issues with a mixture of mixtures of Gaussians in which some param eters are shared to reflect the similar signals observed in different states of the reactor. An EM algorithm for these shared Gaussian mixtu res is presented. Experimental results on nuclear plant data demonstra te the advantages of the proposed approach with respect to the above t wo points.