Research On The Technologies Of Loose Parts Impact Simulation Of NPP And Signal-Noise Separation

As an important source of clean energy,nuclear development has been of great importance to many countries.The installed capability of nuclear power units also keeps increasing gradually.To reduce occurrence probability of serious accidents in nuclear power plants further and ensure the safety of reactors,many safety monitoring systems and engineered safety features have been designed in nuclear power plants.In the history of nuclear power plants,there are many accidents caused by falling parts in the reactor,which eventually resulted in the shutdown of reactor.Thus,in the reactor,loose parts monitoring system are generally installed for monitoring ? analyzing and positioning loose parts.This dissertation focuses on the implications of strong background noise on the loose parts monitoring system in reactor operation process.Based on the existing PWR equipment arrangement,the composition and sensor arrangement of loose parts monitoring system impact signal of loose parts,together with the features of background noise,are analyzed in details;furthermore,the problems of de-noising and signal-noise separation in strong background noise are analyzed further.In the situation of lack of nuclear power plant practical loose parts impact signal,this dissertation applies Hertz collision theory and numerical calculation methods to design a simplified experimental device to simulate impact of loose parts,analyzes and simulates the process of the ball hitting the target.Utilizing acceleration sensor to collect simulant impact signal,and analyzing the features of background noise,generates simulant noise signal according to a certain frequency distribution in MATLAB.Lastly,compounded simulant impact signal and simulated background noise on basis of certain ratio,the mixed signal is used to simulate the signal of loose parts detected in loose parts monitoring system.To achieve the purpose of choosing de-nosing threshold adaptively,wavelet packet analysis technology is used and time window is added to wavelet packet coefficient.Meanwhile,a new threshold processing function is used to execute de-nosing for loose parts impact signal with noise signal.The results indicate that the method proposed can track theintensity change of background noise,suppress background noise effectively,recognize loose parts impact signal accurately and extract relevant features.In order to get complete features of impact signal further,this dissertation also explores the application of blind source separation algorithm in signal-noise Separation of loose parts impact signal.Assuming that background noise and impact noise mix linearly,the kurtosis is used to separate the signal from background noise to get the loose parts impact signal of more complement frequency characteristic.It is the basic of locating and mass estimating for loose parts.