Nuclear power plant performance study by using neural networks

The thermal performance data obtained from the Tennessee Valley Authority (TVA) Sequoyah nuclear power plant show that the heat rate is changing constantly and the plant is probably losing some megawatts of electric power due to variation of the heat rate. It is very difficult to analyze the raw data recorded weekly during the full power operation of the plant because a nuclear power plant is a very complex system with thousands of parameters. The hybrid type of neural network was set up to work as the internal model for prediction of the heat rate. A sensitivity study was then applied to extract information about the key parameters which might strongly affect the plant thermal performance. The preliminary results show that neural networks can be used to analyze plant data and extract some useful information which may be difficult to obtain through traditional analytical models.<>