CN113820741B - Seismic inversion initial model construction method based on deep learning - Google Patents
Seismic inversion initial model construction method based on deep learning Download PDFInfo
- Publication number
- CN113820741B CN113820741B CN202110945639.7A CN202110945639A CN113820741B CN 113820741 B CN113820741 B CN 113820741B CN 202110945639 A CN202110945639 A CN 202110945639A CN 113820741 B CN113820741 B CN 113820741B
- Authority
- CN
- China
- Prior art keywords
- seismic
- data
- attribute
- frequency
- initial model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000010276 construction Methods 0.000 title claims abstract description 19
- 238000013135 deep learning Methods 0.000 title claims abstract description 13
- 238000000034 method Methods 0.000 claims abstract description 26
- 238000013528 artificial neural network Methods 0.000 claims abstract description 12
- 238000004364 calculation method Methods 0.000 claims description 30
- 238000001914 filtration Methods 0.000 claims description 15
- 238000013507 mapping Methods 0.000 claims description 6
- 238000001228 spectrum Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 4
- 230000001052 transient effect Effects 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims 1
- 230000000694 effects Effects 0.000 description 6
- 208000035126 Facies Diseases 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 2
- 239000000047 product Substances 0.000 description 2
- 239000004215 Carbon black (E152) Substances 0.000 description 1
- 239000006227 byproduct Substances 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000007789 gas Substances 0.000 description 1
- 229930195733 hydrocarbon Natural products 0.000 description 1
- 150000002430 hydrocarbons Chemical class 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000003345 natural gas Substances 0.000 description 1
- 239000003209 petroleum derivative Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. for interpretation or for event detection
- G01V1/282—Application of seismic models, synthetic seismograms
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Acoustics & Sound (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention discloses a seismic inversion initial model construction method based on deep learning, which takes a logging longitudinal wave impedance curve, a logging transverse wave impedance curve, a logging longitudinal and transverse wave velocity ratio curve and a logging density curve as learning targets, adopts a deep feedforward neural network algorithm, fuses a conventional inversion initial model and seismic attributes and aims to construct an inversion initial model with low-frequency information and stratum structure information; the initial model constructed by the method is better consistent with the stratum attitude characteristics, the problem of 'time-through' existing in the traditional initial model can be effectively solved, high-density horizon interpretation data are not needed, the workload of horizon interpretation can be greatly reduced, and the inversion result corresponding to the initial model constructed by the method is better consistent with geological knowledge.
Description
Technical Field
The invention belongs to the field of petroleum and natural gas exploration and development, and particularly relates to a seismic inversion initial model construction method based on deep learning.
Background
Seismic inversion is a main means for developing reservoir prediction and hydrocarbon detection, and an initial model required by the inversion is one of key factors influencing the inversion effect. The existing seismic inversion initial model construction method utilizes horizon and fault to construct an isochronal sequence frame, and adopts a proper interpolation algorithm to extrapolate low-frequency information in logging along the sequence frame to obtain an initial model required by inversion. The existing technical scheme for constructing the seismic inversion initial model mainly has the following problems: 1) The prior art scheme is only suitable for sedimentary facies mainly based on vertical product effect, and for sedimentary facies mainly based on side product effect and front product effect, the initial model constructed by the prior art scheme has a 'time-through' phenomenon, so that an inversion result and stratum attitude are contradictory. 2) When a complex fracture system is developed in a research area, the prior art scheme needs high-density horizon interpretation data to ensure the precision of an isochronal sequence frame, so that the interpretation workload and difficulty can be greatly increased, and the seismic inversion progress and effect are influenced.
Disclosure of Invention
In view of the above problems, the present invention provides an inversion initial model construction method based on deep learning. The invention is suitable for the field of seismic data processing of exploration and development of large and medium oil and gas fields under complex geological conditions. Particularly, a logging longitudinal wave impedance curve, a logging transverse wave impedance curve, a logging longitudinal-transverse wave velocity ratio curve and a logging density curve are used as learning targets, a depth feedforward neural network algorithm is adopted, a conventional inversion initial model and seismic attributes are fused, and the purpose is to construct an inversion initial model with low-frequency information and stratum structure information. The initial model constructed by the method is better consistent with the stratum attitude characteristics, the problem of 'time-through' existing in the traditional initial model can be effectively solved, high-density horizon interpretation data are not needed, the workload of horizon interpretation can be greatly reduced, and the inversion result corresponding to the initial model constructed by the method is better consistent with geological knowledge.
The invention is implemented by adopting the following technical scheme:
1. a seismic inversion initial model construction method based on deep learning comprises the following steps:
s1, acquiring logging curve data, explaining horizon data and post-stack seismic data as input data of a subsequent calculation step;
s2, respectively generating longitudinal wave impedance curve data, transverse wave impedance curve data, longitudinal and transverse wave speed ratio curve data and density curve data according to longitudinal wave speed, transverse wave speed and density data in the logging curve data, wherein: longitudinal wave impedance is equal to longitudinal wave velocity multiplied by density, shear wave impedance is equal to shear wave velocity multiplied by density, and the ratio of the longitudinal wave velocity to the shear wave velocity is equal to the longitudinal wave velocity divided by the shear wave velocity;
s3, calculating the explained horizon data and longitudinal wave impedance curve, transverse wave impedance curve, longitudinal and transverse wave velocity ratio curve and density curve data through the existing inversion initial model construction method to generate an original inversion initial model data body; in the existing inversion initial model construction method, a conventional interpolation algorithm, such as an inverse distance square algorithm, is selected, and a logging longitudinal wave impedance curve, a transverse wave impedance curve, a longitudinal and transverse wave velocity ratio curve and a density curve are subjected to spatial interpolation along interpretation horizon data, so that an initial model data volume is obtained. The existing inversion initial model construction method is greatly influenced by the accuracy of the explained horizon data, and under the complex geological condition, the quality of an inversion initial model obtained by the existing inversion initial model construction method is poor due to the fact that the explained horizon data are difficult to pick up accurately and the workload is extremely large;
s4, calculating the post-stack seismic data to obtain a seismic channel integral attribute, a seismic orthogonal channel attribute, a seismic frequency filtering attribute, a seismic main frequency attribute, a seismic average frequency attribute, a seismic polarity attribute, a seismic instantaneous attribute, a seismic envelope attribute, a seismic first derivative attribute and a seismic second derivative attribute, wherein the definition and the geological meaning of the attributes are as follows:
seismic trace integral attribute: this property highlights the mid and low frequency components of the seismic signal and enables a negative 90 degree phase rotation of the seismic data. The calculation method is to perform integral operation on the seismic data to obtain the seismic data.
Seismic frequency filtering properties: the attribute emphasizes information of the seismic signal within a certain frequency band; the calculation method comprises the steps of firstly carrying out Fourier transform on the seismic data, selecting a filtering range in a frequency domain, and then carrying out Fourier inverse transform to obtain the filtered seismic data.
Seismic dominant frequency attributes: the attribute expresses the dominant frequency size of the seismic signal in a certain time window range; selecting a time window range and sliding, performing Fourier transform on the seismic data in each time window range, wherein the frequency corresponding to the maximum value in the amplitude spectrum is the seismic dominant frequency;
seismic average frequency attribute: the attribute expresses the average of the frequency of the seismic signal over a certain time window. The calculation method comprises the steps of selecting a time window range, sliding, carrying out Fourier transform on seismic data in each time window range, and carrying out weighted average on each frequency according to the amplitude spectrum.
Seismic transient attributes: the attributes comprise two attributes of instantaneous frequency and instantaneous phase, wherein the instantaneous frequency attribute expresses the information of the frequency of the seismic signal at a certain moment, and the instantaneous phase attribute expresses the phase information of the seismic signal at a certain moment; the calculation method comprises the following steps of performing Hilbert transform on seismic data, and calculating a plurality of channels obtained after the transform by the following formula:
wherein phi t Is instantaneous phase, w t Is the instantaneous frequency, s t Is the real part after Hilbert transform, h t Is the imaginary part after the hilbert transform and t is time.
Seismic envelope attributes: the attribute represents the intensity of the seismic signal reflection energy. The calculation method comprises the following steps of performing Hilbert transform on seismic data, and calculating the envelope of a complex channel obtained after the transform, wherein the formula is as follows:
wherein A is the seismic envelope, s t Is the real part after Hilbert transform, h t For imaginary part after Hilbert transform
Seismic polarity property: this attribute expresses the apparent polarity of the seismic signal. The calculation method comprises the steps of performing Hilbert transform on seismic data, calculating the envelope of a plurality of channels obtained after the transform, wherein the polarity of a real part of the plurality of channels corresponding to the peak value of the envelope is the seismic apparent polarity.
Seismic orthogonal trace attributes: the attribute expresses the relative wave impedance magnitude of the seismic signal. The calculation method is that Hilbert transform is carried out on the seismic data, and the imaginary part in the complex track obtained after the transform is the orthogonal track attribute.
Seismic first derivative attribute: this property highlights medium and high frequency information in the seismic signal. The calculation method is to calculate the first derivative of the seismic data.
Seismic second derivative attribute: this property highlights high frequency information in the seismic signal. The calculation method is to calculate the second derivative of the seismic data.
S5, extracting various seismic attributes calculated in the step S4 and initial model data in the step S3 at a well point;
s6, selecting different logging curve data as label data of deep learning according to an inversion method;
and S7, taking the tag data as a learning target, and establishing a nonlinear mapping relation between seismic attributes at well points, an original inversion initial model at the well points and a logging tag data body by adopting a depth feedforward neural network algorithm. The deep feedforward neural network comprises an input layer, a plurality of hidden layers and an output layer, and the parameters in the network are optimized and adjusted through a back propagation algorithm. Compared with the traditional feedforward neural network algorithm, the deep feedforward neural network has more hidden layers, can represent more complex functional relationships and realize more effective characteristic representation.
S8, converting the seismic attribute data volume and the original inversion initial model data volume into a label data volume through a nonlinear mapping relation;
and S9, performing low-pass filtering processing on the tag data volume, and filtering out frequency components of 15hz and above to finally obtain an improved inversion initial model. For the post-stack inversion method, an improved longitudinal wave impedance initial model data volume is obtained; for the prestack inversion method, improved longitudinal wave impedance, transverse wave impedance (or longitudinal-transverse wave velocity ratio) and density initial model data volumes are obtained respectively.
Further, the inversion method in S5 includes a post-stack inversion method that selects a longitudinal wave impedance curve as the tag data, and a pre-stack inversion method that selects a longitudinal wave impedance curve, a transverse wave impedance curve, a longitudinal-transverse wave velocity ratio curve, and a density curve as the tag data.
Advantageous effects
1. The inversion initial model constructed by the prior art scheme and the seismic attributes are fused through a deep learning algorithm, and the inversion initial model with low-frequency information and stratum structure information is constructed.
2. The initial model constructed by the method is better consistent with the stratum occurrence characteristics, and the inversion result corresponding to the initial model constructed by the method is more consistent with geological understanding.
3. The method obtains the stratum structure information in the seismic data through the seismic attributes, does not need high-density horizon interpretation data, can greatly reduce the workload of horizon interpretation, and is suitable for constructing the inversion initial model of areas with complex structures and complex stratum structures.
Drawings
FIG. 1 is a schematic flow diagram of the process of the present invention;
FIG. 2 is an inverse initial model constructed in a prior art approach;
FIG. 3 is a label data volume obtained using a deep feed-forward neural network learning algorithm;
FIG. 4 is an inversion initial model constructed according to the technical solution of the present invention;
FIG. 5 is an inversion result corresponding to an initial model based on a prior art approach;
fig. 6 is an inversion result corresponding to the initial model obtained based on the technical solution of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the following detailed discussion of the present invention will be made with reference to the accompanying drawings and examples, which are only illustrative and not limiting, and the scope of the present invention is not limited thereby.
As shown in FIG. 1, the seismic inversion initial model construction method based on deep learning comprises the following steps:
s1, acquiring logging curve data, explaining horizon data and post-stack seismic data as input data of a subsequent calculation step;
s2, extracting longitudinal wave speed, transverse wave speed and density data in the logging curve, calculating a longitudinal wave impedance curve by using the longitudinal wave speed and the density, calculating a transverse wave impedance curve by using the transverse wave speed and the density, and calculating a longitudinal wave speed ratio curve by using the longitudinal wave speed and the transverse wave speed. Wherein, the longitudinal wave impedance is equal to the longitudinal wave velocity multiplied by the density, the transverse wave impedance is equal to the transverse wave velocity multiplied by the density, and the longitudinal-transverse wave velocity ratio is equal to the longitudinal wave velocity divided by the transverse wave velocity;
and S3, combining the layer data with the longitudinal wave impedance curve, the transverse wave impedance curve, the longitudinal-transverse wave velocity ratio curve and the density curve, and calculating a seismic inversion initial model by using the conventional initial model construction method to serve as an original inversion initial model data volume (figure 2). In the existing inversion initial model construction method, a conventional interpolation algorithm, such as an inverse distance square algorithm, is selected, and a logging longitudinal wave impedance curve, a transverse wave impedance curve, a longitudinal and transverse wave velocity ratio curve and a density curve are subjected to spatial interpolation along interpretation horizon data, so that an initial model data volume is obtained. The existing inversion initial model construction method is greatly influenced by the accuracy of the interpreted horizon data, and under the complex geological condition, the quality of an inversion initial model obtained by the existing inversion initial model construction method is poor due to the fact that the interpreted horizon data are difficult to pick up accurately and the workload is extremely large;
s4, calculating a plurality of post-stack seismic data volumes, including a seismic channel integral attribute, a seismic orthogonal channel attribute, a seismic frequency filtering attribute, a seismic main frequency attribute, a seismic average frequency attribute, a seismic polarity attribute, a seismic instantaneous attribute, a seismic envelope attribute, a seismic first derivative attribute and a seismic second derivative attribute, wherein the definition and the geological meaning of the attributes are as follows:
seismic trace integral attribute: this property highlights the mid and low frequency components of the seismic signal and enables a negative 90 degree phase rotation of the seismic data. The calculation method is that the seismic data are obtained by integral operation.
Seismic frequency filtering properties: this attribute highlights the information of the seismic signal within a certain frequency band. The calculation method comprises the steps of firstly carrying out Fourier transform on the seismic data, selecting a filtering range in a frequency domain, and then carrying out Fourier inverse transform to obtain the filtered seismic data.
Seismic dominant frequency attributes: the attribute expresses the dominant frequency magnitude of the seismic signal over a time window. The calculation method comprises the steps of selecting time window ranges and sliding, carrying out Fourier transform on seismic data in each time window range, and obtaining the frequency corresponding to the maximum value in the amplitude spectrum as the seismic main frequency.
Seismic average frequency attribute: the attribute expresses the average of the frequency of the seismic signal over a certain time window. The calculation method comprises the steps of selecting a time window range, sliding, carrying out Fourier transform on seismic data in each time window range, and carrying out weighted average on each frequency according to the amplitude spectrum.
Seismic transient attributes: the attributes comprise two attributes of instantaneous frequency and instantaneous phase, wherein the instantaneous frequency attribute expresses information of the frequency of the seismic signal at a certain moment, and the instantaneous phase attribute expresses phase information of the seismic signal at a certain moment. The calculation method is that Hilbert transform is carried out on the seismic data, and a plurality of channels obtained after the transform are calculated through the following formula:
wherein phi t Is instantaneous phase, w t Is the instantaneous frequency, s t Is the real part after Hilbert transform, h t Is the imaginary part after the hilbert transform and t is time.
Seismic envelope attributes: the attribute represents the intensity of the seismic signal reflection energy. The calculation method comprises the following steps of performing Hilbert transform on seismic data, and calculating the envelope of a complex channel obtained after the transform, wherein the formula is as follows:
wherein A is the seismic envelope, s t Is the real part after Hilbert transform, h t For imaginary part after Hilbert transform
Seismic polarity property: the attribute expresses the apparent polarity of the seismic signal. The calculation method comprises the steps of performing Hilbert transform on seismic data, calculating the envelope of a complex channel obtained after the transform, wherein the polarity of a real part of the complex channel corresponding to the peak value of the envelope is the seismic apparent polarity.
Seismic orthogonal trace attributes: the attribute expresses the relative wave impedance magnitude of the seismic signal. The calculation method is that Hilbert transform is carried out on the seismic data, and imaginary parts in the complex channels obtained after the transform are orthogonal channel attributes.
Seismic first derivative attribute: this property highlights medium and high frequency information in the seismic signal. The calculation method is to calculate the first derivative of the seismic data.
Seismic second derivative attribute: this property highlights high frequency information in the seismic signal. The calculation method is to calculate the second derivative of the seismic data.
S5, extracting various seismic attributes calculated in the step S4 and original inversion initial model data in the step S3 at a well point;
and S6, selecting different well logging curves as label data of deep learning according to different inversion methods. If the method is a post-stack inversion method, selecting a longitudinal wave impedance curve as label data, and if the method is a pre-stack inversion method, selecting a longitudinal wave impedance curve, a transverse wave impedance curve (or a longitudinal-transverse wave velocity ratio curve) and a density curve as label data;
and S7, taking the tag data in the step 6 as a learning target, and establishing a nonlinear mapping relation between seismic attributes at well points, an original inversion initial model at the well points and logging tag data by adopting a depth feedforward neural network algorithm. The deep feedforward neural network comprises an input layer, a plurality of hidden layers and an output layer, and the parameters in the network are optimized and adjusted through a back propagation algorithm. Compared with the traditional feedforward neural network algorithm, the deep feedforward neural network contains more hidden layers, can represent more complex functional relations and realize more effective characteristic representation;
s8, converting the seismic attribute data volume and the basic initial model data volume into a label data volume (figure 3) by using the mapping relation obtained in the step 7;
and S9, performing low-pass filtering processing on the label data volume obtained in the step 8, and filtering out frequency components of 15hz and above to finally obtain an improved inversion initial model (figure 4). Compared with the method based on the prior inversion initial model construction method (figure 5), the inversion result is better matched with geological knowledge when the initial model constructed by the method is used for seismic inversion (figure 6).
The present invention is not limited to the above-described embodiments. The foregoing description of the specific embodiments is intended to describe and illustrate the technical solutions of the present invention, and the above specific embodiments are merely illustrative and not restrictive. Those skilled in the art can make many changes and modifications to the invention without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (2)
1. A seismic inversion initial model construction method based on deep learning is characterized by comprising the following steps:
s1, acquiring logging curve data, explaining horizon data and post-stack seismic data as input data of a subsequent calculation step;
s2, respectively generating longitudinal wave impedance curve data, transverse wave impedance curve data, longitudinal and transverse wave speed ratio curve data and density curve data according to longitudinal wave speed, transverse wave speed and density data in the logging curve data;
s3, calculating the explained horizon data and longitudinal wave impedance curve, transverse wave impedance curve, longitudinal and transverse wave velocity ratio curve and density curve data through the existing inversion initial model construction method to generate an original inversion initial model data volume;
s4, calculating the post-stack seismic data to obtain a seismic channel integral attribute, a seismic orthogonal channel attribute, a seismic frequency filtering attribute, a seismic main frequency attribute, a seismic average frequency attribute, a seismic polarity attribute, a seismic instantaneous attribute, a seismic envelope attribute, a seismic first derivative attribute and a seismic second derivative attribute; wherein:
seismic trace integral attribute: this property highlights the medium and low frequency components of the seismic signal and enables negative 90 degree phase rotation of the seismic data; the calculation method is that the integral operation is carried out on the seismic data to obtain;
seismic frequency filtering properties: the attribute emphasizes information of the seismic signal within a certain frequency band; the calculation method comprises the steps of firstly carrying out Fourier transform on seismic data, selecting a filtering range in a frequency domain, and then carrying out Fourier inverse transform to obtain the filtered seismic data;
seismic dominant frequency attributes: the attribute expresses the dominant frequency of the seismic signal in a certain time window range; selecting a time window range and sliding, performing Fourier transform on the seismic data in each time window range, wherein the frequency corresponding to the maximum value in the amplitude spectrum is the seismic dominant frequency;
seismic average frequency attribute: the attribute expresses the average value of the frequency of the seismic signal in a certain time window range; selecting a time window range and sliding, performing Fourier transform on the seismic data in each time window range, and performing weighted average on each frequency according to the amplitude spectrum;
seismic transient attributes: the attributes comprise two attributes of instantaneous frequency and instantaneous phase, wherein the instantaneous frequency attribute expresses the information of the frequency of the seismic signal at a certain moment, and the instantaneous phase attribute expresses the phase information of the seismic signal at a certain moment; the calculation method is that Hilbert transform is carried out on the seismic data, and a plurality of channels obtained after the transform are calculated through the following formula:
wherein phi t Is instantaneous phase, w t Is the instantaneous frequency, s t Is the real part after Hilbert transform, h t Is the imaginary part after Hilbert transform, and t is time;
seismic envelope attributes: the attribute represents the strength of the reflected energy of the seismic signal; the calculation method comprises the steps of performing Hilbert transform on seismic data, and calculating the envelope of a plurality of channels obtained after the transform, wherein the formula is as follows:
wherein A is the seismic envelope, s t Is the real part after Hilbert transform, h t Is the imaginary part after the hilbert transform;
seismic polarity attributes: the attribute expresses the apparent polarity of the seismic signal; the calculation method comprises the steps of performing Hilbert transform on seismic data, calculating the envelope of a plurality of channels obtained after the transform, wherein the polarity of a real part of the plurality of channels corresponding to the peak value of the envelope is the seismic apparent polarity;
seismic orthogonal trace attributes: the attribute expresses the relative wave impedance magnitude of the seismic signal; the calculation method comprises the steps of carrying out Hilbert transform on seismic data, wherein imaginary parts in a plurality of channels obtained after transformation are orthogonal channel attributes;
seismic first derivative attribute: the attribute emphasizes medium and high frequency information in the seismic signal; the calculation method comprises the steps of solving a first derivative of the seismic data;
seismic second derivative attribute: the attribute emphasizes high frequency information in the seismic signal; the calculation method comprises the steps of solving a second derivative of the seismic data;
s5, extracting various seismic attributes calculated in the step S4 and basic initial model data in the step S3 at a well point;
s6, selecting different logging curve data as label data of deep learning according to an inversion method;
s7, taking the tag data in the step S6 as a learning target, and establishing a nonlinear mapping relation between seismic attributes at well points, a basic initial model at the well points and a logging tag data body by adopting a deep feed-forward neural network algorithm;
s8, converting the seismic attribute data volume and the basic initial model data volume into a label data volume by using the mapping relation obtained in the step S7;
and S9, performing low-pass filtering processing on the label data volume obtained in the step S8, and filtering out frequency components of 15hz and above to finally obtain an improved inversion initial model.
2. The method for constructing the seismic inversion initial model based on the deep learning as claimed in claim 1, wherein the inversion method in S6 includes a post-stack inversion method for selecting a longitudinal wave impedance curve as tag data and a pre-stack inversion method for selecting a longitudinal wave impedance curve, a shear wave impedance curve, a longitudinal-to-transverse wave velocity ratio curve and a density curve as tag data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110945639.7A CN113820741B (en) | 2021-08-16 | 2021-08-16 | Seismic inversion initial model construction method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110945639.7A CN113820741B (en) | 2021-08-16 | 2021-08-16 | Seismic inversion initial model construction method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113820741A CN113820741A (en) | 2021-12-21 |
| CN113820741B true CN113820741B (en) | 2022-11-29 |
Family
ID=78922895
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110945639.7A Active CN113820741B (en) | 2021-08-16 | 2021-08-16 | Seismic inversion initial model construction method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113820741B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116559949B (en) * | 2023-05-19 | 2024-07-12 | 北京宸宇金源科技有限公司 | Carbonate reservoir prediction method, system and equipment based on deep learning |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6982927B2 (en) * | 2003-09-24 | 2006-01-03 | Rdsp I, L.P. | Method for processing seismic data to identify anomalous absorption zones |
| CN105607120A (en) * | 2016-01-19 | 2016-05-25 | 中国海洋石油总公司 | Time-shifting-logging-based method for building initial model with seismic facies constraint |
| WO2019222031A1 (en) * | 2018-05-12 | 2019-11-21 | Schlumberger Technology Corporation | Seismic data interpretation system |
| CA3122686C (en) * | 2018-12-11 | 2023-10-24 | Exxonmobil Upstream Research Company | Automated reservoir modeling using deep generative networks |
| CN109633750A (en) * | 2019-01-18 | 2019-04-16 | 中国地质大学（北京） | Nonlinear mapping relationship inversion method based on well logging phase wave impedance and seismic waveform |
| CN110231653B (en) * | 2019-04-24 | 2020-02-18 | 中海石油(中国)有限公司上海分公司 | Method and device for constructing bidirectional constraint initial model |
| US20210089897A1 (en) * | 2019-09-24 | 2021-03-25 | Quantico Energy Solutions Llc | High-resolution earth modeling using artificial intelligence |
| CN111474576B (en) * | 2020-04-21 | 2022-10-21 | 中海石油(中国)有限公司 | Construction method of pre-stack seismic gather inversion initial model for keeping stratum structure |
| CN112017289B (en) * | 2020-08-31 | 2023-03-24 | 电子科技大学 | Well-seismic combined initial lithology model construction method based on deep learning |
-
2021
- 2021-08-16 CN CN202110945639.7A patent/CN113820741B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN113820741A (en) | 2021-12-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yuan et al. | Double-scale supervised inversion with a data-driven forward model for low-frequency impedance recovery | |
| CN104570067B (en) | Phased seismic inversion method in a kind of geophysical exploration | |
| CN108802812A (en) | Well-seismic fusion stratum lithology inversion method | |
| Yanhu et al. | A method of seismic meme inversion and its application | |
| Du et al. | Seismic facies analysis based on self-organizing map and empirical mode decomposition | |
| CN103389513A (en) | Method for improving seismic data resolution by using acoustic logging data constraint inversion | |
| Zhang et al. | Predicting gas-bearing distribution using DNN based on multi-component seismic data: Quality evaluation using structural and fracture factors | |
| CN108508489B (en) | Seismic inversion method based on waveform micro-variation matching | |
| Jesus et al. | Multiattribute framework analysis for the identification of carbonate mounds in the Brazilian presalt zone | |
| CN104297800A (en) | Self-phase-control prestack inversion method | |
| CN113820741B (en) | Seismic inversion initial model construction method based on deep learning | |
| Gao et al. | Incorporating structural constraint into the machine learning high-resolution seismic reconstruction | |
| Laudon et al. | An enhanced fault detection workflow combining machine learning and seismic attributes yields an improved fault model for Caspian Sea asset | |
| CN110687597A (en) | Wave impedance inversion method based on joint dictionary | |
| Ge et al. | High-resolution seismic impedance inversion integrating the closed-loop convolutional neural network and geostatistics: an application to the thin interbedded reservoir | |
| Kadkhodaie et al. | Pore throat size characterization of carbonate reservoirs by integrating core data, well logs and seismic attributes | |
| CN106291675A (en) | A kind of geological data reconstructing method based on base tracer technique | |
| Melani et al. | The use of variational mode decomposition in assisting sedimentary cyclicity analysis: A case study from an Albian carbonate reservoir, Campos Basin, southeast Brazil | |
| CN105929453B (en) | State estimation method for infinite distribution time lag neural network system with channel fading | |
| CN108508486A (en) | A kind of FRACTURE PREDICTION analysis method based on structural evolution stratum strain | |
| Ningkai et al. | Stepped and detailed seismic prediction of shallow-thin reservoirs in Chunfeng oilfield of Junggar Basin, NW China | |
| Wu et al. | An unsupervised inversion method for seismic brittleness parameters driven by the physical equation | |
| CN113514884A (en) | Compact sandstone reservoir prediction method | |
| Zhang et al. | Broadband reconstruction of seismic signal with generative recurrent adversarial network | |
| Li et al. | An efficient deep learning method for VSP wavefield separation: A DAS-VSP case |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |