NOISE-ANALYSIS METHOD FOR MONITORING THE MODERATOR TEMPERATURE-COEFFICIENT OF PRESSURIZED-WATER REACTORS - NEURAL-NETWORK CALIBRATION

A neural network was trained with data for the frequency, response fun ction between in-core neutron noise and core-exit thermocouple noise i n a pressurized water reactor, with the moderator temperature coeffici ent (MTC) as target. The trained network was subsequently used to pred ict the MTC at other points in the same fuel cycle. Results support us e of the method for operating pressurized water reactors provided nois e data can be accumuluted for several fuel cycles to provide a trainin g base.