Abstract: Using well-log acoustic travel time,natural potential and GR data,the paper reconstructs the travel time of acoustic waves,analyzes the mechanism difference of response of geological body to the three curves,provides correction method for relative depth shift of these curves,and studies the covariance matrix eigenvector filtering approach for de-noising of multiple log curves.On the basis of the above researches,we reconstruct the acoustic travel-time curves based on wavelet multi-scale decomposition.After analyzing the characters of useful signal and noise of wavelet decomposition under various scales with the three curves,we filtered the high frequency part of multi-scale decomposition results of the three curves using correlation filtering technique at adjacent decomposition scales.To ensure the trueness of the reconstructed log curve,we analyzed thoroughly the redundancy(correlativity)and the incorrelativity of the multi-curve multi-scale decomposition results with eigenvalue techniques,and eventually realized the curve reconstruction.Comparison between the reconstructed curve with the drill-hole core log shows an obviously higher resolution of the former in distinguishing sandstone and mudstone.

Key words: multiple log curve, denoising, wavelet analyses, orthogonal analyses, curve reconstruction

摘要： 利用测井声波时差、自然电位和伽玛重构声波时差,分析了以上3种曲线结果对地质体响应方面的机制差异,给出了这些曲线的深度相对移动校正的方法。研究了多条测井曲线去噪的协方差矩阵特征向量滤波方法。在此基础上,利用小波多尺度分解技术进行声波时差曲线重构。在分析这3种测井曲线不同尺度小波分解结果的信号和噪音特点后,利用相邻分解尺度相关滤波技术,对各种曲线的小尺度(高频)分解结果进行滤波处理。为保证重构声波曲线的真实性,深入分析了多曲线、多尺度分解结果的冗余性(相关性)问题;利用特征值技术,对高频多尺度分解分量进行了非相关(正交)分析,最后实现了声波时差曲线的重构,并对重构结果与钻井岩心录井资料进行了对比验证,重构后的声波曲线在区分砂、泥岩性分辨率方面具有明显的提高。

关键词: 多测井曲线, 去噪处理, 小波分析, 正交分析, 曲线重构