中文摘要：

从时间领域导出的松驰时间系列(RTS ) 导致了极化数据(TDIP ) 是有用的估计油水库毛孔结构。然而由于到 signal-to-noise 比率(SNR ) 的敏感，传统的单个价值分解(SVD ) 的倒置精确性有 SNR 的减少的倒置方法还原剂。以便提高倒置精确性并且改进倒置方法的坚韧性到 SNR，基于抑制因素和光谱部件剩余修正，一个改进倒置方法在这研究被建议。RTS 的摆动从 SVD 方法导出的数字倒置结果表演与 SNR 的减少增加了，它使得到精确倒置部件不可能。然而， SNR 几乎没在改进方法的倒置部件上有小影响，并且 RTS 有高倒置精确性和坚韧性。而且，从核心样品数据导出的 RTS 基本上从记载数据的实际导致的极化与毛孔尺寸分发曲线，和 RTS 一致被导出光滑、连续，它显示改进方法是适用的。

英文摘要：

Relaxation time spectra （RTS） derived from time domain induced polarization data （TDIP） are helpful to assess oil reservoir pore structures. However, due to the sensitivity to the signal-to-noise ratio （SNR）, the inversion accuracy of the traditional singular value decomposition （SVD） inversion method reduces with a decrease of SNR. In order to enhance the inversion accuracy and improve robustness of the inversion method to the SNR, an improved inversion method, based on damping factor and spectrum component residual correction, is proposed in this study. The numerical inversion results show that the oscillation of the RTS derived from the SVD method increased with a decrease of SNR, which makes it impossible to get accurate inversion components. However, the SNR has little influence on inversion components of the improved method, and the RTS has high inversion accuracy and robustness. Moreover, RTS derived from core sample data is basically in accord with the pore-size distribution curve, and the RTS derived from the actual induced polarization logging data is smooth and continuous, which indicates that the improved method is practicable.