时频分析方法在提取阵列声波测井信息中的应用
|
摘要
本文在研究前人阵列声波数据处理方法的基础上,提出了用时频分析的方法处理声波数据,时频分析在处理非平稳、多分量信号有着很好的效果,在很多领域得到了应用。对阵列声波信号进行时频分析,能够看出不同振型的波能量、时间和频率的分布特征。本文首先对原始数据进行预处理,进而分别对不同的岩性地层、以及储集层与非储集层的阵列声波数据进行时频分析。重点对不同油水层性质的储集层进行了分析总结,分析其时频特征的共性和区别,利用这种特征分布指导我们区分油水层,最终取得了很好的效果。
Sonic logging is one of the main methods of geophysical logging. Earth's acoustic properties are closely correlated, such as speed, attenuation and the type of rock along with fluid nature. So the earth medium speed and attenuation can provide important information for its medium and changes. Means of measurements also change from the previous monopole logging to S-wave dipole logging, cross dipole anisotropy logging and multi-polar array of acoustic logging. The basic principle of array acoustic logging is to use the large number of repeated information of array to make up for the loss of information caused by a variety of possible measurement errors. Acoustic transducer can receive a variety of acoustic signals. First is the P-wave, followed by S-wave, and then followed by a string of guided waves. Guided Wave is composed of two different modes of wave, one is a so-called pseudo-Rayleigh wave related with the sound energy limited by the wall near and the other is the stoneley wave. Times of different modes arriving at the receiving transducer are different and the different times can be used to analyze the various characteristics of its spread, so as to gain further the characteristics of medium surrounding the well.
The principal means of the array acoustic data processing now are in frequency domain and time domain. But time domain way only comes down to the time domain. The superposition and overlapping caused by the short time information data can create difficulty in processing effective information. Although the Fourier transform of frequency domain methods set up a channel from the time domain to the frequency domain, there was no effective together time domain with frequency domain. When doing Fourier Transform in the entire signal, it only showed the total strength of a certain frequency in the whole signals. Usually it could not provide information related with the time frequency local information of the spectrum component. But in practice, the time information of the spectrum is obtained expectedly. Therefore time-frequency analysis used to process data was proposed in this paper. That was when analyzing signals, relating the information about frequency of acoustic signals with time information. Time-frequency analysis methods could relate the information of the time-domain with that of the frequency domain, through time-frequency analysis of the signal window and get better results.
In this paper, the second-time-frequency analysis—Choi-Williams distribution which had a more obvious physical sense was used. It has gotten very good results for the acoustic array which contained a number of components, non-stationary signals. It admitted index function for the nuclear function and got the time-frequency distribution. In addition, it was also the smoothing window definition of Wigner-Ville distribution so as to make it better curb cross-term and has better resolution.
The major achievements of this paper are as followed:
(1) Analyzing the raw data, eliminate the gaining of the raw signal and Fourier transform filter processing to provide accurate, non- interferential data for the behind the work.
(2) The Choi-Williams energy distribution of the original wave out of the array acoustic, which was obtained by measurements of the typical, different broken rock band encountered by the Chinese Continental Scientific Drilling, was calculated in this paper. Different Lithology corresponding to different time-frequency distribution was also analyzed and summarized.
(3) The Choi-Williams energy distribution picture (contour and 3D) of the array measured by array acoustic logging in the center SongNan depressing structural belt was gained. The different characteristics of the reservoir and non-reservoir were summed up. And then the time-frequency characteristics of the oil-water reservoir with different nature of the oil-water was analyzed and summarized especially.
Major conclusions of this paper are as followed:
(1) Through processing the data of the array acoustic with time-frequency method, noise and other interference signals are not only suppressed, but also distribution characteristics of the energy of P-wave, the S-wave, Stoneley wave and Pseudo-Rayleigh wave etc. at different times, different frequency. P-wave is in the (800μs-1500μs,7kHz-12kHz) around, the S-wave in the (1800μs-2200μs,8kHz-10 kHz), Stoneley wave are in the (2200μs-4000μs,0 kHz-5 kHz), and Pseudo-Rayleigh wave clearly visible, the time is both in the S-wave and the Stoneley wave. The frequency of S-wave is lower than that of P-wave and the frequency of Stoneley wave is lower than that of S-wave. But the frequency of Pseudo-Rayleigh wave is basically from low-frequency to high-frequency. The distribution and energy comparison of various modes of wave can be seen clearly from the energy distribution three-dimensional picture. In general, energy of the P-wave is the smallest, and the S-wave and Stoneley are larger, Pseudo-Rayleigh wave throughout the whole frequency but its energy is very small.
(2) Through making time-frequency characteristics picture of the different lithology broken belts, it can be seen that the broken belt with different lithology has different nature and the characteristics of Choi-Williams energy distribution is significantly different. Its time-frequency distribution of very complicated, and time and frequency of the appearance of P-wave and S-wave in contact broken belt, composed of serpentinite containing pyrite and garnet, are difficult to see. The Stoneley wave (Pseudo-Rayleigh) wave appeared ahead of time, and energy of low-frequency has reduced, but energy of high-frequency has increased. The Pseudo-Rayleigh wave and mud wave followed by appeared accordingly ahead of time, and the increase of duration time and frequency distribution are very different. Its energy intensity in the low-frequency wave corresponded to the Stoneley wave front, but it was stronger than the Stoneley wave front in the high-frequency and the duration time was very long. The appearance time of the appearance of P-wave and S-wave in structure broken belt composed of phengite eclogite was difficult to distinct clearly, and the range (length, width of the frequency) and the intension of duration have increased relatively. The Stoneley wave appeared ahead of time and the duration time increased and frequency distribution was from low frequency to high-frequency. The duration time and frequency distribution of Pseudo-Rayleigh wave and mud wave followed by were quite different. The Choi-Williams energy distribution pictures have fully demo-nstrated the complex relationship in other lithology broken belt. The performance of time - frequency - energy distribution was complex and it was also completely different from two former structure belts.
(3) The characteristics of time-frequency distribution of the mud-rock and the reservoir are significantly different. It fully reflects the impact of the content of the mud. Compared the arrival time of P-wave and S-wave in mud-rock layer with that of the reservoir, there was a significant delay, and with increasing of the clay content, the relative rate of P-wave increased and the arrival time had the postponement trend. The arrival time of the S-wave significantly postponed with the increasing of the clay content. The arrival time of the S-wave and P-wave with the clay content was close. But the rate of Stoneley wave decreased and arrival time postponed with the clay content increasing.
(4) Energy differences is relatively small stem from the three-dimensional picture of the dry layer, energy of the Stoneley wave is the largest ,and delayed ,compared the Choi-Williams distribution characteristics of dry layer with that of oil and water layer. The distribution of S-wave is wider, together with the Stoneley wave, and S-wave appears in an earlier time, usually divided into two sections. The obvious characteristics of other layer is that the Pseudo-Rayleigh wave appears in the wave Stone (3000μs around)and the Pseudo-Rayleigh wave can not develop in the S-wave . This is clearly different with oil-water layer.
(5) Compared time-frequency characteristics of the water layer with that of the oil layer (oil-water layer), a clear distinction between the two was found. The energy attenuation of Stoneley wave, influenced by the impact of the penetration rate and oil and water saturation, is serious, and arrival time is delayed. The energy of S-wave is larger in the oil layer, usually higher than that of Stoneley wave. But in the water layer it is less than or equal to the energy value of Stoneley wave. Seen from the three-dimensional picture there was two energy peaks with same energy in the water layer .But there was usual one peak in the oil layer. In addition, the P-wave magnitude was larger in the water layer, and arrival time is delayed.
Sonic logging is one of the main methods of geophysical logging. Earth's acoustic properties are closely correlated, such as speed, attenuation and the type of rock along with fluid nature. So the earth medium speed and attenuation can provide important information for its medium and changes. Means of measurements also change from the previous monopole logging to S-wave dipole logging, cross dipole anisotropy logging and multi-polar array of acoustic logging. The basic principle of array acoustic logging is to use the large number of repeated information of array to make up for the loss of information caused by a variety of possible measurement errors. Acoustic transducer can receive a variety of acoustic signals. First is the P-wave, followed by S-wave, and then followed by a string of guided waves. Guided Wave is composed of two different modes of wave, one is a so-called pseudo-Rayleigh wave related with the sound energy limited by the wall near and the other is the stoneley wave. Times of different modes arriving at the receiving transducer are different and the different times can be used to analyze the various characteristics of its spread, so as to gain further the characteristics of medium surrounding the well.
The principal means of the array acoustic data processing now are in frequency domain and time domain. But time domain way only comes down to the time domain. The superposition and overlapping caused by the short time information data can create difficulty in processing effective information. Although the Fourier transform of frequency domain methods set up a channel from the time domain to the frequency domain, there was no effective together time domain with frequency domain. When doing Fourier Transform in the entire signal, it only showed the total strength of a certain frequency in the whole signals. Usually it could not provide information related with the time frequency local information of the spectrum component. But in practice, the time information of the spectrum is obtained expectedly. Therefore time-frequency analysis used to process data was proposed in this paper. That was when analyzing signals, relating the information about frequency of acoustic signals with time information. Time-frequency analysis methods could relate the information of the time-domain with that of the frequency domain, through time-frequency analysis of the signal window and get better results.
In this paper, the second-time-frequency analysis—Choi-Williams distribution which had a more obvious physical sense was used. It has gotten very good results for the acoustic array which contained a number of components, non-stationary signals. It admitted index function for the nuclear function and got the time-frequency distribution. In addition, it was also the smoothing window definition of Wigner-Ville distribution so as to make it better curb cross-term and has better resolution.
The major achievements of this paper are as followed:
(1) Analyzing the raw data, eliminate the gaining of the raw signal and Fourier transform filter processing to provide accurate, non- interferential data for the behind the work.
(2) The Choi-Williams energy distribution of the original wave out of the array acoustic, which was obtained by measurements of the typical, different broken rock band encountered by the Chinese Continental Scientific Drilling, was calculated in this paper. Different Lithology corresponding to different time-frequency distribution was also analyzed and summarized.
(3) The Choi-Williams energy distribution picture (contour and 3D) of the array measured by array acoustic logging in the center SongNan depressing structural belt was gained. The different characteristics of the reservoir and non-reservoir were summed up. And then the time-frequency characteristics of the oil-water reservoir with different nature of the oil-water was analyzed and summarized especially.
Major conclusions of this paper are as followed:
(1) Through processing the data of the array acoustic with time-frequency method, noise and other interference signals are not only suppressed, but also distribution characteristics of the energy of P-wave, the S-wave, Stoneley wave and Pseudo-Rayleigh wave etc. at different times, different frequency. P-wave is in the (800μs-1500μs,7kHz-12kHz) around, the S-wave in the (1800μs-2200μs,8kHz-10 kHz), Stoneley wave are in the (2200μs-4000μs,0 kHz-5 kHz), and Pseudo-Rayleigh wave clearly visible, the time is both in the S-wave and the Stoneley wave. The frequency of S-wave is lower than that of P-wave and the frequency of Stoneley wave is lower than that of S-wave. But the frequency of Pseudo-Rayleigh wave is basically from low-frequency to high-frequency. The distribution and energy comparison of various modes of wave can be seen clearly from the energy distribution three-dimensional picture. In general, energy of the P-wave is the smallest, and the S-wave and Stoneley are larger, Pseudo-Rayleigh wave throughout the whole frequency but its energy is very small.
(2) Through making time-frequency characteristics picture of the different lithology broken belts, it can be seen that the broken belt with different lithology has different nature and the characteristics of Choi-Williams energy distribution is significantly different. Its time-frequency distribution of very complicated, and time and frequency of the appearance of P-wave and S-wave in contact broken belt, composed of serpentinite containing pyrite and garnet, are difficult to see. The Stoneley wave (Pseudo-Rayleigh) wave appeared ahead of time, and energy of low-frequency has reduced, but energy of high-frequency has increased. The Pseudo-Rayleigh wave and mud wave followed by appeared accordingly ahead of time, and the increase of duration time and frequency distribution are very different. Its energy intensity in the low-frequency wave corresponded to the Stoneley wave front, but it was stronger than the Stoneley wave front in the high-frequency and the duration time was very long. The appearance time of the appearance of P-wave and S-wave in structure broken belt composed of phengite eclogite was difficult to distinct clearly, and the range (length, width of the frequency) and the intension of duration have increased relatively. The Stoneley wave appeared ahead of time and the duration time increased and frequency distribution was from low frequency to high-frequency. The duration time and frequency distribution of Pseudo-Rayleigh wave and mud wave followed by were quite different. The Choi-Williams energy distribution pictures have fully demo-nstrated the complex relationship in other lithology broken belt. The performance of time - frequency - energy distribution was complex and it was also completely different from two former structure belts.
(3) The characteristics of time-frequency distribution of the mud-rock and the reservoir are significantly different. It fully reflects the impact of the content of the mud. Compared the arrival time of P-wave and S-wave in mud-rock layer with that of the reservoir, there was a significant delay, and with increasing of the clay content, the relative rate of P-wave increased and the arrival time had the postponement trend. The arrival time of the S-wave significantly postponed with the increasing of the clay content. The arrival time of the S-wave and P-wave with the clay content was close. But the rate of Stoneley wave decreased and arrival time postponed with the clay content increasing.
(4) Energy differences is relatively small stem from the three-dimensional picture of the dry layer, energy of the Stoneley wave is the largest ,and delayed ,compared the Choi-Williams distribution characteristics of dry layer with that of oil and water layer. The distribution of S-wave is wider, together with the Stoneley wave, and S-wave appears in an earlier time, usually divided into two sections. The obvious characteristics of other layer is that the Pseudo-Rayleigh wave appears in the wave Stone (3000μs around)and the Pseudo-Rayleigh wave can not develop in the S-wave . This is clearly different with oil-water layer.
(5) Compared time-frequency characteristics of the water layer with that of the oil layer (oil-water layer), a clear distinction between the two was found. The energy attenuation of Stoneley wave, influenced by the impact of the penetration rate and oil and water saturation, is serious, and arrival time is delayed. The energy of S-wave is larger in the oil layer, usually higher than that of Stoneley wave. But in the water layer it is less than or equal to the energy value of Stoneley wave. Seen from the three-dimensional picture there was two energy peaks with same energy in the water layer .But there was usual one peak in the oil layer. In addition, the P-wave magnitude was larger in the water layer, and arrival time is delayed.
引文
[1] 王祝文、刘菁华、聂春燕,基于 Choi-Williams 时频分布的阵列声波测井信号时频分析[J],地球理学进展,2007,22(5), 1481~1486
[2] 唐晓明、郑传汉著,赵晓敏译,定量测井声学[M],北京:石油工业出版社,2004:1～20
[3] 江万哲、章广成,时频分析方法在声波测井信息提取中的应用[J], 石油天然气学报(江汉石油学院学报), 2005, 27(6),736～738
[4] 殷文、印兴耀,基于 MPI 的时频分布的改进及应用[J],地球物理进展, 2005,20(1):165～169
[5] 龚仁荣、顾建祖、骆英、柳祖亭,Gabor 小波时频分析在声发射信号处理中的应用[J],中国测试技术,2006,32(1),76～79
[6] 朱爱民、李斌、陈炜等,时频分析综述及其应用[J],通信与广播电视,2006 年第 4 期:1～7
[7] 吕 菱 、强 智,时频分析在阵列信号处理中的应用[J],现代电子科技,2004,10:63～65
[8] 邹文、陈爱萍、顾汉明,地震信号的时频分析方法[J],世界地质,2004,23(1):95～99
[9] 王西文、杨孔庆、周赢宏等,基于小波变换的地震相干体算法研究[J],地球物理学报,2002,45(6):847～ 853
[10] 张 帆、钟羽云、朱新运等,时频分析方法及在地震波谱研究中的应用[J],地震地磁观测与研究,2006,27(4):17～22
[11] 李舟波,钻井地球物理勘探[M],地质出版社,1986:83～107
[12] 楚泽涵,声波测井原理[M],石油工业出版社,1985:147～152,176～180
[13] 王朝辉,交叉偶极声波测井(XMAC)信息提取方法研究,硕士学位论文,吉林大学,2005
[14] 王朝辉、王祝文,声波测井的发展及现状,地震学研究新发展,石油工业出版社,2004
[15] 吕秀梅,偏心点源发声信号的数值仿真及利用交叉偶极声波测井资料提取地层各向异性信息的方法与应用研究,硕士学位论文,吉林大学,2003
[16] 陈雨红、杨长春、曹齐放等,几种时频分析方法比较[J],地球物理学进展,2006,21(43):1180～1185
[17] 邱天爽、张旭秀、李小兵、孙永梅,统计信号处理[M],北京:电子工业出版社,2004:1～15
[18] 刘波、文忠、曾涯,Matlab 信号处理 [M], 北京:电子工业出版社,2006:126～161
[19] 葛哲学、陈仲生,Matlab 时频分析技术及其应用[M],北京:人民邮电出版社,2006:2～17,66～82
[20] 张学涛、王祝文、原镜海,利用时频分析方法在阵列声波测井中区分油水层[J],岩性油气藏,2008,20(1):101~104
[21] Tang X M and Chunduru R K. Simultaneous inversion of formation shear-wave anisotropy parameters from cross-dipole acoustic-array wave data. Geophysics, 1999, 64(5): 1502~1511
[22] Tang X M and Patterson D. Shear wave Anisotropy Measurement using cross-dipole acoustic logging: An overview. Geophysics, 2001, 42(2):107~117
[23] Andrew L Kurkjian and Shu-Kong Chang. Acoustic multipole sources in fluid-filled boreholes. Geophysics, 1986, 51(1): 148~163
[24] 章成广、王冠贵、黄文新,地层物性与声波全波列波形的关系[J],江汉石油学院学报,1991,13(2):37～45
[25] 楚泽涵、黄贺雄,估算砂泥质岩层孔隙度的新方法——声波全波列测井资料应用研究之一[J],测井技术,1992,16(2):110～116
[26] 田素月、孙灵芬、魏阳庆,声波测井在油藏工程中的应用[J],测井技术,2001,25(5):384～385
[27] 边瑞雪,声波全波列测井中类瑞利波速度及幅度的讨论[J], 石油仪器, 1997, 11(1), 12～ 13
[28] 江明、李兆阳,声波全波列测井资料的处理方法及应用[J],石油仪器,2005,19(5),53～58
[29] 陈必孝、张筠,声波全波列测井资料分析处理技术及应用 [J], 测井技术,2002,26(5):369～372
[30] 沈建国,应用声学基础——实轴积分法及二维谱技术[M],天津大学出版社,2004:64～106
[31] 李太宝,计算声学——声场的方程和计算方法[M],北京:科学出版社,2004:145～180
[32] Leon Cohn. Time-Frequency Analysis: Theory and Applications [M]. Englewood Cliffs: Prentice Hall, 1994
[33] Mallat S,Zhang Z.Matching Pursuits with Time-frequency Dictionaries[J].IEEE Trans.Signal Processing,I993,4I(I2):3397~3380
[2] 唐晓明、郑传汉著,赵晓敏译,定量测井声学[M],北京:石油工业出版社,2004:1～20
[3] 江万哲、章广成,时频分析方法在声波测井信息提取中的应用[J], 石油天然气学报(江汉石油学院学报), 2005, 27(6),736～738
[4] 殷文、印兴耀,基于 MPI 的时频分布的改进及应用[J],地球物理进展, 2005,20(1):165～169
[5] 龚仁荣、顾建祖、骆英、柳祖亭,Gabor 小波时频分析在声发射信号处理中的应用[J],中国测试技术,2006,32(1),76～79
[6] 朱爱民、李斌、陈炜等,时频分析综述及其应用[J],通信与广播电视,2006 年第 4 期:1～7
[7] 吕 菱 、强 智,时频分析在阵列信号处理中的应用[J],现代电子科技,2004,10:63～65
[8] 邹文、陈爱萍、顾汉明,地震信号的时频分析方法[J],世界地质,2004,23(1):95～99
[9] 王西文、杨孔庆、周赢宏等,基于小波变换的地震相干体算法研究[J],地球物理学报,2002,45(6):847～ 853
[10] 张 帆、钟羽云、朱新运等,时频分析方法及在地震波谱研究中的应用[J],地震地磁观测与研究,2006,27(4):17～22
[11] 李舟波,钻井地球物理勘探[M],地质出版社,1986:83～107
[12] 楚泽涵,声波测井原理[M],石油工业出版社,1985:147～152,176～180
[13] 王朝辉,交叉偶极声波测井(XMAC)信息提取方法研究,硕士学位论文,吉林大学,2005
[14] 王朝辉、王祝文,声波测井的发展及现状,地震学研究新发展,石油工业出版社,2004
[15] 吕秀梅,偏心点源发声信号的数值仿真及利用交叉偶极声波测井资料提取地层各向异性信息的方法与应用研究,硕士学位论文,吉林大学,2003
[16] 陈雨红、杨长春、曹齐放等,几种时频分析方法比较[J],地球物理学进展,2006,21(43):1180～1185
[17] 邱天爽、张旭秀、李小兵、孙永梅,统计信号处理[M],北京:电子工业出版社,2004:1～15
[18] 刘波、文忠、曾涯,Matlab 信号处理 [M], 北京:电子工业出版社,2006:126～161
[19] 葛哲学、陈仲生,Matlab 时频分析技术及其应用[M],北京:人民邮电出版社,2006:2～17,66～82
[20] 张学涛、王祝文、原镜海,利用时频分析方法在阵列声波测井中区分油水层[J],岩性油气藏,2008,20(1):101~104
[21] Tang X M and Chunduru R K. Simultaneous inversion of formation shear-wave anisotropy parameters from cross-dipole acoustic-array wave data. Geophysics, 1999, 64(5): 1502~1511
[22] Tang X M and Patterson D. Shear wave Anisotropy Measurement using cross-dipole acoustic logging: An overview. Geophysics, 2001, 42(2):107~117
[23] Andrew L Kurkjian and Shu-Kong Chang. Acoustic multipole sources in fluid-filled boreholes. Geophysics, 1986, 51(1): 148~163
[24] 章成广、王冠贵、黄文新,地层物性与声波全波列波形的关系[J],江汉石油学院学报,1991,13(2):37～45
[25] 楚泽涵、黄贺雄,估算砂泥质岩层孔隙度的新方法——声波全波列测井资料应用研究之一[J],测井技术,1992,16(2):110～116
[26] 田素月、孙灵芬、魏阳庆,声波测井在油藏工程中的应用[J],测井技术,2001,25(5):384～385
[27] 边瑞雪,声波全波列测井中类瑞利波速度及幅度的讨论[J], 石油仪器, 1997, 11(1), 12～ 13
[28] 江明、李兆阳,声波全波列测井资料的处理方法及应用[J],石油仪器,2005,19(5),53～58
[29] 陈必孝、张筠,声波全波列测井资料分析处理技术及应用 [J], 测井技术,2002,26(5):369～372
[30] 沈建国,应用声学基础——实轴积分法及二维谱技术[M],天津大学出版社,2004:64～106
[31] 李太宝,计算声学——声场的方程和计算方法[M],北京:科学出版社,2004:145～180
[32] Leon Cohn. Time-Frequency Analysis: Theory and Applications [M]. Englewood Cliffs: Prentice Hall, 1994
[33] Mallat S,Zhang Z.Matching Pursuits with Time-frequency Dictionaries[J].IEEE Trans.Signal Processing,I993,4I(I2):3397~3380