A reservoir prediction method and system based on multi-source information fusion

Abstract

This invention discloses a reservoir prediction method and system based on multi-source information fusion, belonging to the field of reservoir prediction technology. The method includes: high-fidelity signal processing of seismic data combined with seismic geological conditions of the target area; fine well-seismic calibration using well logging, seismic, and geological information combined with domain knowledge; seismic facies identification using a deep embedded self-organizing map network; multi-scale fault detection using cepstral analysis-enhanced coherent ant body attributes; wave impedance and porosity inversion using seismic waveform indication inversion; reservoir gas-bearing capacity detection by combining data-driven seismic texture analysis, depth domain dispersion analysis, and deep learning of seismic data; and prediction of favorable reservoir distribution by adaptively weighted fusion of the results of seismic facies identification, fault detection, wave impedance and porosity inversion, and gas-bearing capacity detection. The advantage of this invention is that it can achieve accurate and reliable prediction of deep and complex oil and gas reservoirs by integrating multi-source, multi-dimensional, and multi-attribute data.

Description

Technical Field

[0001] This invention relates to the fields of seismic data processing and reservoir prediction technology, and in particular to a reservoir prediction method and system based on multi-source information fusion. Background Technology

[0002] The goal of reservoir prediction is to determine the spatial distribution and reserves of oil and gas reservoirs, making it one of the most crucial aspects of oil and gas exploration. With the continuous growth of global demand for oil and gas resources and the increasing depth of exploration, the targets of oil and gas exploration are gradually shifting from shallow to medium-deep and even deep reservoirs, from simple to complex, and from conventional to unconventional. Reservoir prediction has become an effective means to improve the accuracy of deep oil and gas reservoir characterization and reduce exploration risks.

[0003] Reservoir prediction is a comprehensive research endeavor aimed at quantitatively characterizing the spatial distribution, physical parameters, and fluid distribution of subsurface reservoirs through the integrated analysis of multi-source, multi-scale data. Current technologies typically employ single geophysical attributes or independent, staged analysis models, which are insufficient to meet the requirements for detailed characterization of deep, complex oil and gas reservoirs. Therefore, there is an urgent need to develop more effective reservoir prediction methods to achieve the coordinated characterization of reservoir spatial distribution, physical parameters, and fluid properties. Summary of the Invention

[0004] To address the aforementioned shortcomings in existing technologies, this invention provides a reservoir prediction method and system based on multi-source information fusion to solve the problem of coordinated characterization of reservoir spatial distribution, physical parameters, and fluid properties, thereby achieving stable and high-precision prediction of deep and complex reservoirs and providing a reliable foundation for efficient exploration and development of deep and complex gas reservoirs.

[0005] To achieve the above objectives and technical effects, the present invention provides the following technical solution.

[0006] According to one aspect of the present invention, a reservoir prediction method based on multi-source information fusion is provided, comprising the following steps:

[0007] (1) Acquire data on the target area, including well logging, coring, seismic and geological data, and perform data analysis and preprocessing;

[0008] (2) Combine the seismic geological conditions of the target area to perform high-fidelity and weak signal processing on the seismic data to obtain high-fidelity, high signal-to-noise ratio and high-resolution seismic data;

[0009] (3) Use information from well logging, seismic and geological data, combined with knowledge from the fields of geology and geophysics, to perform fine well seismic calibration and accurately calibrate the target layer;

[0010] (4) A deep embedded self-organizing mapping network that can adaptively consider the changes in formation thickness is used for seismic facies visualization analysis, and the sedimentary facies type and distribution characteristics of the reservoir are determined by combining well logging and core data;

[0011] (5) Perform cepstral analysis on seismic data to enhance the seismic response characteristics of fractures, calculate coherence and ant body properties by combining gradient structure tensor algorithm, and perform multi-scale fracture detection by combining FMI imaging logging interpretation;

[0012] (6) Use seismic waveform inversion to obtain parameters such as density and wave impedance that are closely related to rock porosity, convert porosity according to the statistical relationship of well data, and combine rock physical analysis to divide a relatively large area of ​​favorable blocks;

[0013] (7) Use data-driven seismic texture analysis, depth domain dispersion analysis, and deep learning of seismic data to extract seismic attribute features that are sensitive to gas content, and then fuse the calculated attribute features according to the proportion of the dominant part to detect the gas content of the reservoir.

[0014] (8) Predict favorable reservoir distribution by using the results of adaptive weighted fusion seismic facies identification, multi-scale fracture detection, wave impedance and porosity inversion and gas content detection;

[0015] (9) Analyze and evaluate the favorable reservoir distribution results obtained by fusion prediction of multi-source, multi-dimensional and multi-attribute data based on the well logging interpretation of the drilled wells in the target area and the gas test and production data of single wells.

[0016] Preferably, the data preprocessing in step (1) may include outlier removal, environmental correction, multi-well consistency correction, etc.

[0017] Preferably, the process of high-fidelity weak signal processing of seismic data in step (2) may include static correction processing combining tomographic static correction and residual static correction of reflected waves, broadband high-fidelity weak signal noise suppression processing, pre-stack high-fidelity amplitude compensation and high-resolution processing, etc., with the purpose of weak signal protection considered in each step.

[0018] Preferably, the fine well seismic calibration described in step (3) is based on the conventional well seismic calibration process, and uses prior geological information and knowledge in the fields of geology and geophysics to constrain and obtain results that are more consistent with the actual geological conditions of the target area, so as to accurately establish the connection between depth domain geological, logging and other information and time domain seismic geophysical information.

[0019] Preferably, the deep embedded self-organizing map network described in step (4) is a seismic facies visualization analysis method that simultaneously completes the self-encoding feature extraction process and self-organizing map feature clustering. Furthermore, by adding sparse constraints, the stability of the seismic facies classification results is improved, thereby enabling the visualization results to display more details.

[0020] Preferably, the cepstral analysis described in step (5) is performed by performing spectral analysis on the seismic signal and taking the logarithm of the spectrum to obtain the cepstral coefficients, thereby approximately quantitatively characterizing the low-frequency enhancement effect of the seismic phase axis and enhancing the seismic signal of small faults.

[0021] Preferably, the seismic waveform indication inversion described in step (6) uses the lateral variation of the seismic waveform to replace the spatial domain interpolation simulation of the variogram function, thereby realizing the inversion under automatic control of seismic facies. This can overcome the subjectivity caused by the requirement of human pre-determining the sedimentary facies in traditional facies inversion, and is a true facies inversion.

[0022] Preferably, the seismic pattern analysis described in step (7) is a seismic data analysis and seismic geological information extraction method developed by drawing on speech feature extraction methods to identify reservoir pore fluid characteristics from seismic signals, which can extract weak seismic response characteristics of reservoir pore fluid.

[0023] Preferably, the depth domain dispersion analysis mentioned in step (7) refers to performing seismic dispersion analysis in the depth domain rather than the time domain. The depth domain can better reflect the actual location and content changes of gas content, including: establishing a high-precision velocity model of the target area, converting the time domain seismic data from the time domain to the depth domain according to the velocity model, and extracting the seismic dispersion attenuation attribute in the depth domain.

[0024] Preferably, the deep learning of seismic data in step (7) utilizes the nonlinear representation and knowledge discovery capabilities of deep neural networks to mine and utilize the weak and insignificant seismic geological information hidden in the seismic data.

[0025] Preferably, the purpose of the adaptive weighted fusion in step (8) is to fuse and unify the obtained multi-source, multi-dimensional, and multi-attribute data to achieve comprehensive prediction of favorable reservoirs.

[0026] According to another aspect of the present invention, a reservoir prediction system based on multi-source information fusion is provided. The system includes a processor and a memory, the memory storing a computer program executable on the processor. When the processor executes the computer program, it implements a reservoir prediction method based on multi-source information fusion. The system includes:

[0027] The data acquisition module is used to acquire data including well logging, coring, seismic and geological data, and to perform data analysis and preprocessing.

[0028] The high-fidelity and low-fidelity signal processing module is used to perform high-fidelity and low-fidelity signal processing on seismic data based on the seismic geological conditions of the target area.

[0029] The well-seismic fine calibration module is used to perform fine well-seismic calibration based on logging, seismic, and geological information, combined with domain knowledge.

[0030] The seismic facies analysis module is used to perform deep embedded self-organizing mapping seismic facies visualization analysis on seismic data, and to determine the sedimentary facies type and distribution characteristics of reservoirs by combining well logging and core data;

[0031] The fracture detection module is used to perform cepstral analysis on seismic data to enhance the seismic response characteristics of fractures, calculate coherence and ant-body properties by combining gradient structure tensor algorithm, and perform multi-scale fracture detection by combining FMI imaging logging interpretation.

[0032] The porosity inversion module is used to perform seismic waveform indication inversion on seismic data to obtain parameters such as density and wave impedance that are closely related to rock porosity. Based on the statistical relationship of well data, porosity is obtained by conversion, and combined with rock physical analysis, a relatively large area of ​​favorable blocks is delineated.

[0033] The gas-bearing detection module is used to extract gas-sensitive seismic attribute features from seismic data through data-driven seismic texture analysis, depth domain dispersion analysis, and deep learning of seismic data. The calculated attribute features are then fused according to the proportion of the dominant components to detect the gas-bearing capacity of the reservoir.

[0034] The reservoir comprehensive prediction module is used to predict favorable reservoir distribution by adaptively weighted fusion of seismic facies identification, multi-scale fault detection, wave impedance and porosity inversion and gas-bearing detection results.

[0035] The prediction result evaluation module is used to analyze and evaluate the favorable reservoir distribution results obtained by fusing multi-source, multi-dimensional, and multi-attribute data based on well logging interpretations of drilled wells in the target area, as well as gas testing and production data from single wells.

[0036] Compared with existing technologies, the beneficial effects of this invention are as follows: Based on high-fidelity and weak signal processing of seismic data and fine well seismic calibration, this invention organically combines seismic facies visualization analysis that can adaptively consider formation thickness variations, multi-scale fracture detection of coherent ant body attributes enhanced by cepstral analysis, wave impedance and porosity inversion based on seismic waveform indication, and multi-dimensional reservoir gas-bearing detection combining seismic texture analysis, depth domain dispersion analysis, and deep learning of seismic data. This forms a reservoir prediction method of "facies-zone determination, porosity-reservoir determination, and gas-buoyancy determination". It can overcome the limitations of current technology systems that usually use a single geophysical attribute or a staged independent analysis mode for reservoir prediction, and realize the synergistic characterization of reservoir spatial distribution, physical parameters, and fluid properties, providing a reliable basis for guiding the subsequent development of reservoirs. Attached Figure Description

[0037] Figure 1A flowchart illustrating a reservoir prediction method based on multi-source information fusion, provided for embodiments of this application;

[0038] Figure 2 A schematic diagram of the structure of a reservoir prediction system based on multi-source information fusion provided in this application embodiment;

[0039] Figure 3 This is a map showing the predicted distribution of favorable reservoirs in a target area, provided as an embodiment of this application. Detailed Implementation

[0040] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the specification, and not all embodiments.

[0041] This invention provides a reservoir prediction method based on multi-source information fusion, such as... Figure 1 As shown, in some embodiments, the method includes the following steps:

[0042] Step 101: Obtain data from the target area, including well logging, coring, seismic and geological data, and perform data analysis and preprocessing;

[0043] It should be noted that in some embodiments, the data preprocessing described in step 101 may include outlier removal, environmental correction, multi-well consistency correction, etc.

[0044] Step 102: Combine the seismic geological conditions of the target area to perform high-fidelity and weak signal processing on the seismic data to obtain high-fidelity, high signal-to-noise ratio, and high-resolution seismic data;

[0045] It should be noted that in some embodiments, the process of high-fidelity weak signal processing of seismic data described in step 102 may include static correction processing combining tomographic static correction and residual static correction of reflected waves, broadband high-fidelity weak signal noise suppression processing, pre-stack high-fidelity amplitude compensation and high-resolution processing, etc., with the purpose of weak signal protection considered in each step.

[0046] Step 103: Utilize information from well logging, seismic and geological data, combined with knowledge from the fields of geology and geophysics, to perform fine well-seismic calibration and accurately calibrate the target layer;

[0047] It should be noted that in some embodiments, the fine well seismic calibration described in step 103 is based on the conventional well seismic calibration process, and uses prior geological information and knowledge in the fields of geology and geophysics to constrain and obtain results that are more consistent with the actual geological conditions of the target area, so as to accurately establish the connection between depth domain geological, logging and other information and time domain seismic geophysical information.

[0048] Step 104: Seismic facies visualization analysis is performed using a deep embedded self-organizing mapping network that can adaptively consider changes in formation thickness, and the sedimentary facies type and distribution characteristics of the reservoir are determined by combining well logging and core data.

[0049] It should be noted that in some embodiments, the deep embedded self-organizing map network described in step 104 is a seismic facies visualization analysis method that simultaneously completes the autoencoder feature extraction process and the self-organizing map feature clustering. Furthermore, by adding sparse constraints, the stability of the seismic facies classification results is improved, thereby enabling the visualization results to display more details.

[0050] Step 105: Perform cepstral analysis on the seismic data to enhance the seismic response characteristics of the fractures, calculate the coherence and ant-body properties using the gradient structure tensor algorithm, and perform multi-scale fracture detection using FMI imaging logging interpretation.

[0051] It should be noted that in some embodiments, the cepstral analysis described in step 105 is performed by performing spectral analysis on the seismic signal and taking the logarithm of the spectrum to obtain the cepstral coefficients, thereby approximately quantitatively characterizing features such as the low-frequency enhancement effect of the seismic phase axis and enhancing the seismic signal of small faults.

[0052] Step 106: Use seismic waveform inversion to obtain parameters such as density and wave impedance that are closely related to rock porosity. Based on the statistical relationship of well data, porosity is obtained by conversion. Combined with rock physical analysis, a relatively large area of ​​favorable blocks is delineated.

[0053] It should be noted that in some embodiments, the seismic waveform indication inversion described in step 106 uses the lateral variation of the seismic waveform to replace the spatial domain interpolation simulation of the variogram function, thereby realizing the inversion under automatic control of seismic facies. This can overcome the subjectivity caused by the requirement of human pre-determining the sedimentary facies in traditional facies inversion, and is a true facies inversion.

[0054] Step 107: Extract gas-sensitive seismic attribute features using data-driven seismic texture analysis, depth domain dispersion analysis, and deep learning of seismic data, and then fuse the calculated attribute features according to the proportion of the dominant part to detect reservoir gas content.

[0055] It should be noted that in some embodiments, the seismic pattern analysis described in step 107 is a seismic data analysis and seismic geological information extraction method developed by drawing on speech feature extraction methods to identify reservoir pore fluid characteristics from seismic signals, which can extract weak seismic response characteristics of reservoir pore fluids.

[0056] It should be noted that in some embodiments, the depth domain dispersion analysis described in step 107 refers to performing seismic dispersion analysis in the depth domain rather than the time domain. The depth domain is better able to reflect the actual location and content changes of gas content, including: establishing a high-precision velocity model of the target area, converting time-domain seismic data from the time domain to the depth domain according to the velocity model, and extracting seismic dispersion attenuation attributes in the depth domain.

[0057] It should be noted that in some embodiments, the deep learning of seismic data described in step 107 utilizes the nonlinear representation and knowledge discovery capabilities of deep neural networks to mine and utilize the weak and insignificant seismic geological information hidden in the seismic data.

[0058] Step 108: Predict favorable reservoir distribution by using the results of adaptive weighted fusion seismic facies identification, multi-scale fracture detection, wave impedance and porosity inversion, and gas content detection.

[0059] It should be noted that, in some embodiments, the purpose of the adaptive weighted fusion described in step 108 is to fuse and unify the obtained multi-source, multi-dimensional, and multi-attribute data to achieve comprehensive prediction of favorable reservoirs.

[0060] Step 109: Analyze and evaluate the favorable reservoir distribution results obtained by fusing multi-source, multi-dimensional, and multi-attribute data based on the well logging interpretation of drilled wells in the target area and the gas testing and production data of single wells.

[0061] Corresponding to the above-mentioned reservoir prediction method based on multi-source information fusion, this embodiment of the invention provides a reservoir prediction system based on multi-source information fusion, such as... Figure 2 As shown. In some embodiments, the prediction system includes a processor and a memory, the memory storing a computer program that can run on the processor, the processor executing the computer program to implement the above-described reservoir prediction method based on multi-source information fusion, the system comprising:

[0062] Data acquisition module 201 is used to acquire data including well logging, coring, seismic and geological data, and to perform data analysis and preprocessing.

[0063] The high-fidelity and low-fidelity signal processing module 202 is used to perform high-fidelity and low-fidelity signal processing on seismic data according to the seismic geological conditions of the target area.

[0064] The well-seismic fine calibration module 203 is used to perform fine well-seismic calibration based on logging, seismic and geological information and in combination with domain knowledge.

[0065] The seismic facies analysis module 204 is used to perform deep embedded self-organizing mapping seismic facies visualization analysis on seismic data, and to determine the sedimentary facies type and distribution characteristics of the reservoir by combining well logging and core data;

[0066] The fracture detection module 205 is used to perform cepstral analysis on seismic data to enhance the seismic response characteristics of fractures, calculate coherence and ant-body properties by combining gradient structure tensor algorithm, and perform multi-scale fracture detection by combining FMI imaging logging interpretation.

[0067] The porosity inversion module 206 is used to perform seismic waveform indication inversion on seismic data to obtain parameters such as density and wave impedance that are closely related to rock porosity. Based on the statistical relationship of well data, porosity is obtained by conversion, and a relatively large favorable block is delineated by combining rock physical analysis.

[0068] The gas-bearing detection module 207 is used to extract gas-bearing sensitive seismic attribute features from seismic data through data-driven seismic texture analysis, depth domain dispersion analysis, and seismic data deep learning, and to fuse the calculated attribute features according to the proportion of the dominant part for reservoir gas-bearing detection.

[0069] The reservoir comprehensive prediction module 208 is used to predict favorable reservoir distribution by adaptively weighted fusion of seismic facies identification, multi-scale fault detection, wave impedance and porosity inversion and gas-bearing detection results.

[0070] The prediction result evaluation module 209 is used to analyze and evaluate the favorable reservoir distribution results obtained by fusing multi-source, multi-dimensional, and multi-attribute data based on the well logging interpretation of drilled wells in the target area and the gas testing and production data of single wells.

[0071] Figure 3 This figure shows the predicted distribution of favorable reservoirs in a target area using this technology. As can be seen from the figure, the predicted results conform to the geological characteristics of the target area, reflect the distribution of favorable reservoirs, and are largely consistent with the actual drilling results from well locations not used in the calculations. Overall, the predictions demonstrate high accuracy and reliability.

[0072] Those skilled in the art should understand that the above embodiments are merely illustrative of the beneficial effects of the present invention and are not exhaustive. Any modifications, equivalent changes, improvements, etc., made without departing from the scope and spirit of the described embodiments should not be excluded from the protection scope of the present invention.

Claims

1. A reservoir prediction method based on multi-source information fusion, characterized in that, Includes the following steps: (1) Acquire data from the target area, including well logging, coring, seismic and geological data, and perform data analysis and preprocessing; (2) Combine the seismic geological conditions of the target area to perform high-fidelity and weak signal processing on the seismic data to obtain high-fidelity, high signal-to-noise ratio and high-resolution seismic data; (3) Use well logging, seismic and geological information and combine them with knowledge of geology and geophysics to perform fine well seismic calibration and accurately calibrate the target layer; (4) A deep embedded self-organizing mapping network that can adaptively consider the changes in formation thickness is used for seismic facies visualization analysis, and the sedimentary facies type and distribution characteristics of the reservoir are determined by combining well logging and core data; (5) Perform cepstral analysis on seismic data to enhance the seismic response characteristics of fractures, combine gradient structure tensor algorithm to calculate coherence and ant body properties, and combine formation microresistivity scanning imaging logging interpretation to perform multi-scale fracture detection. (6) The density and wave impedance parameters closely related to rock porosity are obtained by seismic waveform inversion, and porosity is obtained by conversion based on the statistical relationship of well data. The range of favorable blocks is divided in combination with rock physical analysis. (7) Use data-driven seismic texture analysis, depth domain dispersion analysis, and deep learning of seismic data to extract seismic attribute features that are sensitive to gas content, and then fuse the calculated attribute features according to the proportion of the dominant part to detect the gas content of the reservoir. (8) Predict favorable reservoir distribution by using the results of adaptive weighted fusion seismic facies identification, multi-scale fracture detection, wave impedance and porosity inversion and gas-bearing detection; (9) Analyze and evaluate the favorable reservoir distribution results obtained by fusion prediction of multi-source, multi-dimensional and multi-attribute data based on the well logging interpretation of the drilled wells in the target area and the gas test and production data of single wells.

2. The reservoir prediction method based on multi-source information fusion according to claim 1, characterized in that: The process of high-fidelity weak signal processing of seismic data described in step (2) includes static correction processing that combines tomographic static correction and residual static correction of reflected waves, broadband high-fidelity weak signal noise suppression processing, pre-stack high-fidelity amplitude compensation and high-resolution processing steps, with the purpose of weak signal protection considered in each step.

3. The reservoir prediction method based on multi-source information fusion according to claim 1, characterized in that: The fine well seismic calibration described in step (3) is based on the conventional well seismic calibration process. It uses prior geological information and knowledge in the fields of geology and geophysics to constrain the results to obtain results that are more consistent with the actual geological conditions of the target area, and accurately establishes the connection between depth domain geology, well logging information and time domain seismic geophysical information.

4. The reservoir prediction method based on multi-source information fusion according to claim 1, characterized in that: The deep embedded self-organizing map network described in step (4) is a seismic facies visualization analysis method that simultaneously completes the autoencoder feature extraction process and the self-organizing map feature clustering. It also improves the stability of the seismic facies classification results by adding sparse constraints, thereby enabling the visualization results to display more details.

5. The reservoir prediction method based on multi-source information fusion according to claim 1, characterized in that: The seismic pattern analysis described in step (7) is a seismic data analysis and seismic geological information extraction method developed by drawing on speech feature extraction methods to identify reservoir pore fluid characteristics from seismic signals. It can extract the weak seismic response characteristics of reservoir pore fluids.

6. The reservoir prediction method based on multi-source information fusion according to claim 1, characterized in that: The depth domain dispersion analysis mentioned in step (7) refers to performing seismic dispersion analysis in the depth domain rather than the time domain. The depth domain can better reflect the actual location and content changes of gas content. This includes: establishing a high-precision velocity model of the target area, converting time domain seismic data from the time domain to the depth domain according to the velocity model, and extracting seismic dispersion attributes in the depth domain.

7. A reservoir prediction system based on multi-source information fusion, characterized in that, The system includes a processor and a memory, the memory storing a computer program that can run on the processor, and the processor executing the computer program to implement the steps of the reservoir prediction method based on multi-source information fusion as described in claim 1.

Images