Development of the Early Fault Warning System for Key Indicators of Nuclear Power Plant Equipment Based on Neural Network
In order to avoid sudden equipment failures affecting the safety and economy of nuclear power plants,a key index early warning system is developed based on the historical operation data and neural network algorithms.Using the system,key performance of the equipment can be predicted and deviation between the predicted value and the actual measured value can be analyzed by using a variety of control charts,and finally early warning for deviations exceeding the limits can be generated.After the system is deployed in a nuclear power plant,it has successfully realized several early warnings of failures.Take a sudden rise in condenser pressure triggering a DCS alarm event as an example,the system has detected the abnormal rise in condenser pressure and issued an early warning about 4 hours before the DCS alarm and before the inspection by the operation personnel in the main control room,which can reserve enough time for the operation and maintenance personnel to handle the problem.The system can greatly reduce the pressure on the operation personnel to monitor the panels and improves the operation safety and economy of the unit.
NETL
NSTL
万方数据
