Acoustic Leak Detection Based on Wavelet Packet Decomposition for LMFBR Steam Generators

牛晓东,杨献勇
DOI: https://doi.org/10.3321/j.issn:1000-0054.2002.12.025
2002-01-01
Abstract:The detection of water/steam leaks in liquid metal fast breeder reactors (LMFBR) is an important part of the nuclear power plant safety system. This paper describes the application of signal processing methods using wavelet packet transforms to extract the key features from an acoustic leak detection system using signal identification. Two diagnostic methods developed to recognize the leak are based on the energy spectrum and the coefficient matrix based on wavelet packet decomposition. These methods were used to analyze experimental data from the sodium loop in the Liquid Metal Lab in Tsinghua University. The results show that the frequency band of the sodiumwater reaction is less than 6 kHz and that there are distinct differences between normal signals and fault signals. The methods effectively improve the sensing ability of the acoustic leak detection system from 200 mL·h-1 to 10 mL·h-1. 
What problem does this paper attempt to address?