Inversion Of NMR Logging Based On Singular Value Decomposition Method

Nuclear magnetic resonance(NMR)logging technology is a technology developed in recent decades,which is of great significance to the development of logging technology.Compared with the traditional logging technology,NMR logging technology not only has the ideal observation signal-hydrogen nucleus,but also the original signal of the survey is not affected by the rock skeleton,it is the only logging methods which porosity information that can provide the formation fluid,and can provide better permeability,movable fluid,bound water,oil and gas content and other information,suitable for low porosity and low permeability and complex special lithologic oil and gas model,it overcomes the defect of conventional logging methods.In the complex oil and gas exploration in the world,many times are the use of NMR logging to solve the low porosity low permeability,complex special lithology,bound water,heavy oil and other evaluation problems.Based on the basic principles of NMR,relaxation process and the NMR properties of fluids,this paper shudies the inversion process of one-dimensional and two-dimensional NMR logging.The inversion process is the core of the NMR logging,In this paper,the simulated echo signals and the actual echo signals are respectively deduced by singular value decomposition,and the singular value decomposition method introduces the SNR to cut off the singular value,so that the inversion method can automatically adjust the morbidity of the matrix according to the signal-to-noise ratio,and obtain a more stable solution,also on the inversion process of the non-negative restrictions to ensure that the inversion results to meet the actual requirements.This paper studies the one-dimensional NMR logging T2 spectrum distribution,and because of the overlapping of oil and gas in the one-dimensional T2 spectral distribution,and the current study is aimed at one-dimensional NMR logging,therefore,the T2-D spectrum of two-dimensional NMR logging is studied emphatically.The original echo signal of NMR logging has a large error in the inversion result when the SNR is low,So this article on this basis,this paper proposes the method of wavelet transform to denoise the original echo signal,and the experiments show that the inversion effect of the analog signal and the actual signal after noise reduction can be improved better than other studies..The influence of the factors such as the number of singular values,the number of cloth points and the parameters of the signal acquisition on the inversion process is also studied,in the process of signal acquisition and inversion,the different values will have different effects on the inversion results,For the two-dimensional model studied in this paper,it is possible to improve the accuracy and speed of inversion by preserving the number of singular values,the number of points and the short echo interval and high magnetic field gradient.