全波测井速度频散提取方法性能对比与应用

Dispersion information extraction methods for full waveform acoustic logging:Applications and relative performances

全波列数据包含多种频散波,准确有效地提取频散信息可以为储层及含油气性评价等提供更丰富的信息。详细对比了傅里叶变换算法(FTM)、加权频谱相干法(WSS)、振幅相位估计法(APES)、前后向振幅相位估计法(FBAPES)和最小方差无失真波束形成算法(Capon)及前后向最小方差无失真波束形成算法(FBCapon)共6种频散信息提取方法的性能与应用效果。首先,基于已知函数的波形信号与实测数据的处理分析,对比了上述6种算法的保幅性、分辨率及抗噪性。其次,结合实测与模拟声波波列数据,对比了算法的实际应用效果,实现了模式波的分离。最后,考察了基于波场分离频散提取方法的应用效果。分析结果表明,在6种算法中,加权频谱相干算法及傅里叶变换算法抗噪性好,但保幅性能差且分辨率偏低;振幅相位估计类算法振幅估计准确、分辨率较高,但抗噪性较差;最小方差无失真波束形成类算法分辨率、抗噪性以及慢度估计准确性等综合性能较好,但估计的振幅低于正常值。实测声波波列数据处理结果表明,基于长短时方差比值法的时域滤波与频域滤波的波场分离能够有效分离模式波且不改变其速度及频散特征,与综合性能较好的前后向最小方差无失真波束形成算法结合能较好地提取频散信息。

Full-wave column data represent a variety of dispersion waves,and the accurate extraction of dispersion information enables the effective evaluation of reservoir properties and the potential for oil and gas extraction.This study compares the performance and application of six dispersion information extraction methods:the Fourier transform method(FTM),weighted spectral semblance(WSS),amplitude and phase estimation(APES)and its forward-backward(FB)form,and minimum covariance and non-distortion Capon beamforming and its FB form.Firstly,the amplitude preservation of the six algorithms is investigated by designing models using waveform signals with known functions.The anti-noise ability of the algorithms is compared with the slowness estimation error under different noise intensities,and their resolutions are compared based on the processing results of the measured data.Secondly,combined with the measured and simulated data,the practical application of the six algorithms to separating mode waves is compared.Finally,the application of the dispersion extraction methods is examined based on wavefield separation.Among the six algorithms,WSS and FTM exhibit good noise immunity,but large side flaps,poor amplitude preservation,and low resolution;the two APES methods exhibit accurate amplitude estimation and high resolution,but poor noise immunity;and the Capon algorithms have a combination of high resolution,noise immunity,and accuracy of slowness estimation,but an estimated magnitude lower than the real value.The noise immunity of the FB algorithms is higher than the forward-only algorithms,and the FB Capon magnitude estimation error is half that of the forward-only Capon algorithm.The results of the simulated and measured data show that the combination of frequency domain filtering and time domain filtering based on the long to short time variance ratio method can effectively separate mode waves without changing their velocity or dispersion characteristics.Wavefield separation can reduce signal interference and improve the accuracy of dispersion information extraction,such that single-mode analysis can be used to extract dispersion information from multi-mode waves.The combination of wavefield separation and the FB Capon method can even more accurately extract dispersion information.