Nuclear magnetic resonance(NMR)logging can provide a lot of important information for oil and gas reservoir evaluation,the key to extract these information is the inversion of the echo data recorded underground by the NMR logging.The quality of the inversion is directly related to the accuracy of the oil and gas reservoir evaluation.The now available NMR inversion methods give bad resolution for the short relaxation corresponding to the small pores of unconventional reservoir.Therefore it is indispensable to study and advance inversion methods and improve the NMR logging application for unconventional reservoir.One of the most significant problem in the NMR data inversion procedure is the huge memory and time cost resulted from the processing of hundreds of thousands of echo data.So it is an essential issue to compress the raw echo data before the inversion procedure.Based on the existing data compression methods,a new method called principal component analysis(PCA)is proposed and applied to compress 1D and 2D NMR data before the inversion procedure,the results demonstrate that PCA compression method can achieve fast and accurate data compression.In this paper,linear integral transforms are adopted to calculate the petrophysical parameters which are important to evaluate the gas and oil reservoir instead of obtaining the T2 distribution in the first place,which will result in a more accurate solution than that from the traditional method.In addition,another important application of linear integral transform is that the parameters can be used as constraints for inversion.In this paper,these parameters and their uncertainties information are combined to improve the traditional Tikhonov regularization inversion.Compared to the traditional Inverse Laplace Transform inversion method,this method can further narrow the solution down into a more accurate interval,improving the resolution of T2 spectrum and especially the short relaxation part,which is very important for the unconventional reservoir evaluation. |