A core internal vibration monitoring system which is particularly concerned on the core support barrel (CSB) in the nuclear power plant reactor vessel is developed in this work. The core or fuel damage accidents can be caused by the loose-jointed flange between the top of the CSB and the head of the vessel. The loose-jointed flange can be detected with the internal vibration monitoring system, which has conventionally used the signals from ex-core neutron detectors. In order to improve the accuracy of the CSB monitoring system, however, the signals from the piezoelectric accelerometers are used in this work. This thesis consists of two parts; one is the development of a suitable tool for detecting the hold down spring broken accident or wearing out of the CSB, and the other the generation of vibration signals to represent the abnormal states of CSB.
In this thesis, the adaptive resonance theory (ART; a type of neural network) is used to develop the monitoring system. The monitoring system using the Fuzzy ARTMAP processes the signals from the accelerometers. On the other hand, in order to get the data sets of the CSB in abnormal (loose-jointed) states, the finite element method (FEM) is used to model the CSB in various loose-jointed states. The target CSB is the one which is placed in ULJIN nuclear power plant unit 1. A mock-up CSB is constructed and experiments are carried out to prove that the FEM analyses properly simulate the CSB frequency responses in various states. The results show that the CSB FEM analyses and mock-up experiments are in good agreement.