Font Size: a A A

Study Of Extraction Method For Proton MRS Characteristic Parameters

Posted on:2009-01-02Degree:MasterType:Thesis
Country:ChinaCandidate:Q Q SunFull Text:PDF
GTID:2120360242480658Subject:Power electronics and electric drive
Abstract/Summary:PDF Full Text Request
NMR technology is widely used in biology, chemistry, physics ,medicine and other fields, while it has also been effectively used in the field of earth sciences. The surface Nuclear Magnetic Resonance technology for ground water is the latest application of Nuclear Magnetic Resonance technology in the field of earth sciences, and now it is the only direct and effective method of detecting groundwater in the world. Its data processing includes two aspects: First, the preprocessing of NMR data– the extraction of NMR characteristic parameters. Second, the inversion of NMR data. The characteristic parameters of initial amplitude, average decay time and initial phase reflected the size of water in the underground aquifer and the permeability of underground media. The accuracy of NMR characteristic parameters extraction limits the credibility of data inversion.The characteristic parameters of initial amplitude, average decay Time can be directly extracted by fitting the envelope curve of NMR signal. The characteristic parameter of initial phase can not be directly extracted. Faced with such a situation, France NUMIS system artificially produced two reference signals whose frequency equal to the local Larmor frequency, and took orthogonal decomposition of the NMR signal. Through these two orthogonal components, the characteristic parameter of initial phase can be indirectly extracted. Unlike France Orthogonal analysis, a new overall approach of characteristic parameter extraction - Hilbert transform method is presented in this paper.The amplitude of NMR signal is very small, about the level of nv, and the natural electromagnetic background noise is much larger than the NMR signal. Measured signal is real NMR signal or noise needs to be judged firstly. The main frequency of NMR equals to the local Larmor frequency. Therefore, the main frequency of signal needs to be extracted before data processing. Discrete Fourier Transform is the most frequently used method of the spectrum analysis, but there are some errors, including frequency alias, frequency leakage, and picket fence effect. The general theory of frequency leakage concern reducing the overall spectrum error,but NMR spectral analysis only concern the extraction of main frequency, so the general theory of decreasing frequency leakage is inappropriate at here. In view of this situation, new ideas about reducing frequency leakage error are discussed in this paper. In many documents, when sampling time is fixed, the frequency resolution is fixed. To increase the frequency resolution, sampling time must increase. So there is a conclusion that if the frequency resolution is less than 1HZ, the sampling time must be more than 1S. With the increased sampling time, the efficiency of system will reduce. A idea that in some definition the frequency resolution did not necessarily link to sampling time is found in this paper, and views that the frequency resolution is not restricted by sampling time are put.Filter processing, as well as envelope fitting is important steps of NMR data preprocessing. The filtering effect between FIR filter and IIR filter in dealing with NMR signal, and the infection of group delay for NMR envelope distortion are compared in this paper, and had a conclusion that the effect of FIR filter restraining the NMR background noise is better. In the fitting process of NMR envelope, faced with the situation that the error of the least-squares method for non-Gaussian noise is larger. Other fitting methods, including the least absolute deviation method are introduced in this paper, the maximin method, and the mathematical models of them were established, the corresponding calculation methods of them were studied, summed up the suited situation for different fitting methods.Finally, the measured field data processed, extraction of the parameters, using inversion software had the water histogram of the vertical distribution, and compared with the result of France NUMIS system. The method of this paper is suitable, could accurately judge whether the acquisition signals is true NMR signal, the initial amplitude and the average decay time can be extracted accurately, the initial phase can be extracted. So it can be used in the preprocessing software of NMR equipment for water.
Keywords/Search Tags:NMR, characteristic parameters, Hilbert transform method, Orthogonal analysis method, spectrum analysis, curve fitting
PDF Full Text Request
Related items