Initiating event classification and scenario prediction for severe accident management using neural networks

It will be very difficult for nuclear power plant operators to predict and identify the major severe accident scenarios following accident initiation by staring at temporal trends of important parameters. Therefore, in this work, the accidents occurred are classified into categorized initiating events such as Loss Of Coolant Accidents (LOCA), Total Loss Of FeedWater (TLOFW), Station BlackOut (SBO), and Steam Generator Tube Rupture (SGTR) by using a probabilistic neural network that can well be applied to the classification problems. Also, their major severe accident scenarios are identified by fuzzy neural networks using the initial measured signals related to the reactor coolant system, the Steam Generators (S/G), and the containment environments. The fuzzy neural network is a fuzzy inference system equipped with a training algorithm. An automatic structure constructor for the fuzzy neural networks selects the input variables automatically, and optimizes the rule number and its related parameters. It is verified that the proposed algorithm classifies very well initiating events and identifies accurately important time points representing major severe accident scenarios and also break sizes