An NMR log echo data de-noising method based on the wavelet packet threshold algorithm. (26th October 2015)

• Record Type: Journal Article
• Authors: Meng, Xiangning; Xie, Ranhong; Li, Changxi; Hu, Falong; Li, Chaoliu; Zhou, Cancan
• Is Part Of: Journal of geophysics and engineering. Volume 12:Number 6(2015:Dec.)
• Page Start: 956
• Page End: 968
• Publication Date: 2015-10-26
• Subjects: NMR log signal -- wavelet packet threshold algorithm -- SNR improvement -- wavelet packet basis -- decomposition scale; Geophysics -- Periodicals; Prospecting -- Geophysical methods -- Periodicals; Engineering -- Periodicals; 622.1505
• DOI: 10.1088/1742-2132/12/6/956
• Languages: English
• ISSNs: 1742-2132

Abstract

Abstract: To improve the de-noising effects of low signal-to-noise ratio (SNR) nuclear magnetic resonance (NMR) log echo data, this paper applies the wavelet packet threshold algorithm to the data. The principle of the algorithm is elaborated in detail. By comparing the properties of a series of wavelet packet bases and the relevance between them and the NMR log echo train signal, 'sym7' is found to be the optimal wavelet packet basis of the wavelet packet threshold algorithm to de-noise the NMR log echo train signal. A new method is presented to determine the optimal wavelet packet decomposition scale; this is within the scope of its maximum, using the modulus maxima and the Shannon entropy minimum standards to determine the global and local optimal wavelet packet decomposition scales, respectively. The results of applying the method to the simulated and actual NMR log echo data indicate that compared with the wavelet threshold algorithm, the wavelet packet threshold algorithm, which shows higher decomposition accuracy and better de-noising effect, is much more suitable for de-noising low SNR–NMR log echo data.