Prediction of Major Transient Scenarios for Severe Accidents of Nuclear Power Plants Using GMDH

It is important for operators and technical staffs to find out what an initiating event of a severe accident is by observing initial short time trends of major parameters in order to effectively accomplish severe accident management strategies. If they know some time sequences of major severe accident scenarios, they can cope with situations that can lead to the worst situation that plant safety systems do not work appropriately. The present work aims to classify the initiating events which may lead to severe accident conditions such as loss of coolant accidents (LOCA) and steam generator tube rupture (SGTR) by applying a support vector classification (SVC). Also, another objective is to identify the major severe accident scenarios such as core uncovery, core exit temperature (CET) and reactor vessel (RV) failure by using a group method of data handling (GMDH) algorithm of which the inputs are the time-integrated values of important measured signals during a short time interval after reactor scram. The proposed algorithm for the initiating event classification and the major severe accident scenario identification is verified by comparison with the simulation data of the MAAP4 code for the advanced power reactor 1400 (APR1400) developed by Korea Hydro and Nuclear Power company (KHNP)