A Pattern Recognition Method for Nuclear Reactor Core Surveillance

An automatic method for reactor core surveillance is described which employs pattern recognition techniques. A core vibration surveillance method for a pressurized water reactor has been developed which provides the early detection of failures. The method allows on-line application and is able to give quick surveillance decisions (in about one minute). The sensors are eight ex-core ionization chambers. The system is very flexible and adaptable to the statistical properties of the data. The algorithms used are heuristic but need only the definition of a small number of parameters. These parameters may be optimized either by automatic procedures or by explicit experimental methods. Some experimental results are presented.