Using a neural network for abnormal event identification in BWRs

Information on anomalies such as abnormal events is considered to be important for operation support when choosing information to be offered to operators. The authors have applied neural network techniques to identify an abnormal event that causes a reactor scram in boiling water reactors. A primary feature of the method is that the result of the neural network is confirmed using the knowledge base on plant status when each event occurs. This improves the result's reliability. A second feature is that the neural network uses analog data such as reactor pressure, the acquisition of which is triggered by the scram signal. The event identification method is shown. The event identification method is tested using a workstation