Method and device for establishing initial model of seismic inversion based on lithofacies probability inverse mapping

Abstract

The application relates to a method and device for establishing an initial model of seismic inversion based on inverse mapping of lithofacies probability in the field of oil and gas exploration. The method for establishing an initial model of seismic inversion based on inverse mapping of lithofacies probability comprises the following steps: collecting and processing data to obtain post-stack seismic data and a lithofacies curve; obtaining a plurality of seismic attribute bodies from the post-stack seismic data, and combining the lithofacies curve to obtain a pseudo well curve; training a neural network model by using the pseudo well curve, and outputting a lithofacies classification result and a lithofacies probability body; analyzing the lithofacies classification result and the lithofacies probability body to establish a posterior probability density space of physical property parameters; and based on the posterior probability density space and a probability density distribution, using a Monte Carlo algorithm and a rejection-acceptance sampling to generate a physical property parameter combination meeting a condition, and obtaining the initial model of seismic inversion. The application can more accurately perform reservoir characterization, and the consistency between the initial model of seismic inversion and a well point is good.

Description

Technical Field

[0001] This invention relates to the field of oil and gas exploration, and in particular to a method and apparatus for establishing an initial seismic inversion model based on lithofacies probability inverse mapping. Background Technology

[0002] Inverting elastic or physical parameters using pre-stack seismic data is a crucial step in oil and gas exploration and development. Its purpose is to provide a reliable basis for predicting and evaluating oil and gas reservoirs through accurate estimation of subsurface structures and physical parameters. Accurate elastic parameter inversion helps explorers identify potential oil and gas reservoirs, optimize well location design, reduce exploration risks, and improve the development efficiency of oil and gas fields.

[0003] In seismic inversion, the construction of the initial model is crucial, especially as it has a decisive impact on the inversion results. The core of seismic inversion is to match observed seismic data with forward modeling data through iterative optimization to infer the physical parameters of the subsurface medium. However, neglecting lateral variations in strata can lead to insufficient model accuracy, severely affecting the quality of subsurface structural imaging and impacting the accuracy and reliability of the inversion results. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and apparatus for establishing an initial model for seismic inversion based on lithofacies probability inverse mapping.

[0005] The technical solution adopted by this invention to solve its technical problem is: a method for establishing an initial seismic inversion model based on lithofacies probability inverse mapping, comprising the following steps:

[0006] S1. Collect and process oil testing data, seismic data, well logging data and stratigraphic data to obtain post-stack seismic data and lithofacies curves;

[0007] S2. Obtain several seismic attribute volumes from the post-stack seismic data, and combine them with the lithofacies curves to obtain the pseudo-well curves for each seismic attribute volume;

[0008] S3. The neural network model is trained using the pseudo-well curve. The trained neural network model is then combined with the seismic data to output the lithofacies classification results and lithofacies probability volume.

[0009] S4. Analyze the lithofacies classification results and the lithofacies probability volume to obtain the probability density distribution of different physical property parameters and establish the posterior probability density space of the physical property parameters.

[0010] S5. Based on the posterior probability density space and the probability density distribution, traverse the seismic data, and use the Monte Carlo algorithm and rejection-acceptance sampling to generate a combination of physical property parameters that meet the conditions, thereby obtaining the initial seismic inversion model.

[0011] Preferably, step S1 includes:

[0012] Collect the oil testing data, seismic data, well logging data, and stratigraphic data within the work area;

[0013] The seismic data, well logging data, and stratigraphic data are preprocessed to obtain the time-depth relationship after the well-seismic event; the distribution of physical property parameters is obtained through the time-depth relationship.

[0014] Based on the distribution of the physical property parameters, the lithofacies of the oil well are divided to obtain the lithofacies curve of the well.

[0015] The seismic data is then offset and stacked to obtain the post-stacked seismic data.

[0016] Preferably, step S2 includes:

[0017] Based on the post-stack seismic data, the seismic attribute volume is obtained according to preset conditions;

[0018] Based on the inherent resolution of the post-stack seismic data, the resolution of the lithofacies curve is matched with the inherent resolution using a scale coarsening method to obtain the pseudo-well curve.

[0019] Preferably, the seismic attribute body includes an instantaneous amplitude attribute body, an instantaneous frequency attribute body, and a sweet spot attribute body.

[0020] Preferably, step S3 includes:

[0021] Obtain the pseudo-well curves of different seismic attribute bodies at the oil well locations to obtain the training set;

[0022] The neural network model is trained using a supervised learning algorithm and the training set; the trained neural network model is used to classify the seismic data to obtain the lithofacies classification result; the lithofacies classification result includes several lithofacies classifications; each lithofacies classification is calculated using an inverse distance weighted algorithm to obtain the lithofacies probability volume.

[0023] Preferably, step S4 includes:

[0024] Based on the lithofacies classification results, a rock physics model is established, well curves are obtained, and the target parameters of the initial seismic inversion model are determined.

[0025] Based on the lithofacies curves and the target parameters, statistical rock physics analysis is performed on different lithofacies to obtain the histogram distribution and probability density distribution of different physical property parameters in different lithofacies.

[0026] Based on the histogram distribution and the probability density distribution, the posterior probability density space is established.

[0027] Preferably, the probability density distribution includes an N-dimensional probability density distribution;

[0028] The posterior probability density space includes an N-dimensional space of physical property parameters, covering the effective value range of N physical property parameters;

[0029] Where N is the number of physical property parameters involved in the statistical rock physics analysis.

[0030] Preferably, step S5 includes:

[0031] Based on the posterior probability density space, according to the probability density distribution, the posterior probability of different lithofacies corresponding to each point in the posterior probability density space for the combination of physical property parameters is calculated and normalized.

[0032] By iterating through the posterior probability of each sampling point in the seismic data, the lithofacies probability ratio of each sampling point in the seismic data is obtained.

[0033] Based on the Monte Carlo algorithm and probability density function, M combinations of physical property parameters that conform to the preset parameter distribution are generated;

[0034] Based on the rejection-acceptance sampling algorithm and the lithofacies probability ratio of the current sampling point in the seismic data, m combinations of physical parameters that meet the conditions are selected from the M combinations of physical parameters within a preset tolerance.

[0035] The combination of the m physical property parameters is used as the initial model for direct inversion of physical property parameters.

[0036] Preferably, each sampling point in the seismic data includes every point recorded in the seismic trace, with a sampling rate of 2 ms.

[0037] An apparatus for establishing an initial seismic inversion model based on inverse lithofacies probability mapping, the apparatus comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method for establishing an initial seismic inversion model based on inverse lithofacies probability mapping as described above.

[0038] The implementation of this invention has the following beneficial effects:

[0039] This invention, based on well testing data, seismic data, well logging data, and stratigraphic data, acquires post-stack seismic data and lithofacies curves for seismic inversion to obtain pseudo-well curves. These are then combined with a neural network model to obtain lithofacies classification results and lithofacies probability volumes. Further analysis is performed to establish a posterior probability density space for physical property parameters. Finally, Monte Carlo algorithms and rejection-acceptance sampling are used to generate suitable combinations of physical property parameters, resulting in the initial seismic inversion model. This invention, through lithofacies probability inverse mapping, enables more accurate reservoir characterization, resulting in good consistency between the initial seismic inversion model and the reservoir characterization at the well location. Furthermore, the spatial variation of parameters conforms to the actual lithofacies distribution, making it more geologically significant. Attached Figure Description

[0040] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0041] Figure 1 This is a flowchart of a method for establishing an initial seismic inversion model based on lithofacies probability inverse mapping in one embodiment;

[0042] Figure 2 This is a two-dimensional probability density analysis result between in-phase and different physical property parameters in one embodiment;

[0043] Figure 3 This represents the probability density difference of different rock facies under a certain combination of physical property parameters in one embodiment;

[0044] Figure 4 This represents a probability volume for different lithofacies in one embodiment;

[0045] Figure 5 This is an initial seismic inversion model for porosity obtained by inverse mapping of lithofacies probability in one embodiment;

[0046] Figure 6 This is an initial seismic inversion model of pore structure obtained by probabilistic inverse mapping of lithofacies in one embodiment;

[0047] Figure 7 This is an initial model of physical property parameters (porosity) established using conventional interpolation methods;

[0048] Figure 8 This is an initial seismic inversion model (porosity) established according to the lithofacies probability inverse mapping method in one embodiment. Detailed Implementation

[0049] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0050] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0051] The method for establishing an initial seismic inversion model based on lithofacies probability inverse mapping provided in this invention includes the following steps:

[0052] S1. Collect and process oil testing data, seismic data, well logging data and stratigraphic data to obtain post-stack seismic data and lithofacies curves.

[0053] The collection and processing of well testing data, seismic data, well logging data, and stratigraphic data aims to obtain post-stack seismic data and lithofacies curves. Well testing data processing includes analyzing basic well data and well testing results. Seismic data processing utilizes deconvolution, stacking, and migration imaging techniques to improve resolution and signal-to-noise ratio, yielding post-stack seismic data. Well logging data processing involves data cleaning and correction to address geological issues. Stratigraphic data processing accurately determines the stratigraphy through comprehensive calibration. Ultimately, these data are used to generate lithofacies curves, providing crucial geological information for oil and gas exploration.

[0054] S2. Obtain several seismic attribute volumes from the post-stack seismic data, and combine them with lithofacies curves to obtain the pseudo-well curve for each seismic attribute volume.

[0055] First, key attribute volumes, such as amplitude and frequency, are extracted from seismic data using seismic attribute analysis techniques. These attributes reflect the subsurface geological structure and rock physical properties. Second, lithofacies curves from well logging data are used. These curves display the physical properties of rocks, such as porosity and permeability, providing geological evidence for the seismic attribute volumes. Finally, seismic trace matching technology is used to combine the seismic attribute volumes and lithofacies curves to generate corresponding pseudo-well curves, ensuring the accuracy and reliability of the seismic attribute volumes.

[0056] S3. Use pseudo-well curves to train the neural network model, and use the trained neural network model in combination with seismic data to output lithofacies classification results and lithofacies probability volumes.

[0057] First, a neural network is trained using pseudo-well curves, enabling the model to learn the relationship between seismic signals and lithofacies characteristics. Second, the model extracts seismic data features and performs deep learning to capture the spatiotemporal characteristics of the data. Finally, the model outputs lithofacies classifications and probability volumes, providing the likelihood of lithofacies occurrence and offering decision support for oil and gas exploration. By integrating seismic data processing and machine learning techniques, the accuracy and efficiency of lithofacies identification are improved.

[0058] S4. Analyze the lithofacies classification results and lithofacies probability volumes to obtain the probability density distribution of different physical property parameters and establish the posterior probability density space of the physical property parameters.

[0059] The analysis of lithofacies classification results and lithofacies probability volumes involves establishing a lithofacies probability distribution model. This is achieved by loading a three-dimensional structural model and importing lithofacies data, combined with acoustic impedance data, to determine the probability values ​​of the lithofacies data. Using a Bayesian theoretical framework and the Metropolis algorithm, combined with prior geostatistical information, physical property parameters are inverted to obtain the posterior probability density space.

[0060] Furthermore, frameworks and algorithms such as Hamiltonian Monte Carlo method, Markov chain Monte Carlo sampling, ensemble learning method, Bayesian network, multi-point geostatistical simulation method, SNESIM algorithm, GENESIM algorithm, backpropagation neural network, support vector machine regression (R-SVM), multiple regression analysis (MRA), support vector machine classification (C-SVM), and Bayesian stepwise discriminant analysis (BAYSD) can be used to obtain the posterior probability density space.

[0061] S5. Based on the posterior probability density space and probability density distribution, traverse the seismic data, use the Monte Carlo algorithm and rejection-acceptance sampling to generate a combination of physical property parameters that meet the conditions, and obtain the initial model for seismic inversion.

[0062] This method extracts suitable combinations of physical parameters from seismic data using the Monte Carlo algorithm and rejection-acceptance sampling. The Monte Carlo method simulates the distribution of subsurface rock physical parameters through random sampling, while rejection-acceptance sampling selects samples from a proposed distribution to ensure it conforms to the target distribution. This process combines seismic data with geostatistical characteristics to generate possible combinations of physical parameters, constructing an initial subsurface rock physical property model and providing a starting point for seismic inversion. This method effectively handles high-dimensional, nonlinear problems, improving inversion accuracy and efficiency.

[0063] Specifically, the initial seismic inversion model is a dataset. By displaying the initial seismic inversion model graphically, the subsurface structure and rock properties can be visually illustrated, thereby enabling better analysis of geological structure and the distribution of geological resources.

[0064] This invention combines statistical rock physics analysis and seismic lithofacies identification results based on machine learning, and applies the method of probabilistic inverse mapping of lithofacies to allocate different physical property parameters corresponding to different lithofacies. This adds lateral variation information of strata to the initial seismic inversion model and realizes the establishment of a high-precision initial seismic inversion model.

[0065] Specifically, probabilistic inverse facies mapping is a method for constructing initial reservoir facies models. This process combines facies classification information with probabilistic statistical techniques to infer the distribution of elastic and physical parameters corresponding to each facies type, thereby constructing a physical model of the reservoir. This method can more accurately characterize the spatial distribution and uncertainties of reservoir facies, and provide a reliable initial model for seismic data interpretation and reservoir assessment, thus improving the accuracy of reservoir feature identification and characterization.

[0066] In some executable embodiments, step S1 includes:

[0067] Collect oil testing data, seismic data, well logging data, and stratigraphic data within the work area.

[0068] A "work area" refers to a specific geographical region for oil and gas exploration and development, encompassing oil and gas accumulations within a single trap. It is both a reservoir assemblage with a similar geological structure and a specific unit for management and operation. Oil and gas reservoirs within a work area share the same pressure system, and a specialized team is responsible for their exploration, development, and production activities. Collecting various data and information at the work area level facilitates comprehensive analysis of oil, gas, and water well production dynamics, reducing the problem of information silos.

[0069] Furthermore, it facilitates data sharing, replication, and merging, optimizes seismic data acquisition and interpretation, and improves the accuracy of velocity field modeling. Fine calibration of joint well-seismic data enhances the accuracy of stratigraphic interpretation. Big data governance and intelligent logging interpretation applications reduce reliance on expert experience and simplify data processing workflows. The construction of a unified logging database enables integrated data application, improving data security and access efficiency. In addition, standardized processing of seismic attributes and improved accuracy of inter-well reservoir prediction provide strong data support for reservoir management and decision-making, reducing exploration risks, saving costs, and enhancing the scientific rigor and effectiveness of oil and gas exploration.

[0070] Seismic data, well logging data, and stratigraphic data are preprocessed to obtain the time-depth relationship after the seismic event. The distribution of physical property parameters is then obtained from the time-depth relationship.

[0071] Specifically, well logging data and stratigraphic data are transformed from depth-based descriptive data into time-based data. This allows for the correlation between well logging data and stratigraphic data and seismic data, further revealing the distribution of physical parameters.

[0072] Understandably, preprocessing seismic, well logging, and stratigraphic data can significantly improve data quality, ensure consistency, and accurately establish post-seismic time-depth relationships. This process removes noise and outliers, improves data accuracy, and enhances the precision of seismic and tectonic interpretation. Accurate time-depth relationships are crucial for seismic data interpretation, especially in clastic sedimentary formations, significantly improving the accuracy and precision of well-seismic calibration. Furthermore, preprocessed data is more suitable for geological analysis and reservoir evaluation, enhancing data usability, promoting seismic-geological integration, and improving work efficiency. Automated preprocessing workflows accelerate data processing, providing a solid data foundation for seismic inversion and reservoir modeling, thereby improving the efficiency and success rate of oil and gas exploration.

[0073] Based on the distribution of physical property parameters, the lithofacies of the oil wells in the work area are divided to obtain the lithofacies curves of the wells.

[0074] Specifically, based on the well test data report and the distribution of physical properties such as porosity, water saturation, and clay content, the lithofacies are rationally classified to obtain the lithofacies curve for the well. Well operations refer to on-site work and activities directly related to the oil well.

[0075] Furthermore, physical properties refer to the physical characteristics of rocks, such as porosity, permeability, density, and acoustic velocity. The distribution of these parameters reflects how the physical properties of underground rocks vary spatially. Lithofacies refers to rock units with the same or similar petrological characteristics. Through the analysis of physical properties, rocks beneath oil wells can be divided into different lithofacies, such as sandstone, mudstone, and limestone.

[0076] The seismic data is migrated and stacked to obtain post-stack seismic data.

[0077] In some executable embodiments, step S2 includes:

[0078] Based on post-stack seismic data, seismic attribute volumes are obtained according to preset conditions.

[0079] Specifically, seismic attributes such as instantaneous amplitude, instantaneous frequency, relative impedance, trace integral, sweet spot, and frequency division are extracted based on post-stack seismic data.

[0080] Based on the inherent resolution of the post-stack seismic data, the resolution of the lithofacies curves is matched with the inherent resolution using a scale coarsening method to obtain pseudo-well curves.

[0081] It's important to note that the inherent resolution of seismic data refers to the ability of the seismic data itself to resolve the smallest details or structures. This is limited by various factors such as the frequency of the seismic waves, the signal-to-noise ratio of the seismic data, and geological conditions. Scale coarsening can be achieved through various mathematical methods, such as the Backus equivalent averaging method, which processes small-scale well logging curves to make them comparable at a seismic scale. Lithofacies curves, obtained from well logging data, reflect the changes in the physical properties of downhole rocks with depth. The resolution of lithofacies curves refers to their ability to resolve the smallest changes in rock physical properties.

[0082] Furthermore, lithofacies curves refer to lithofacies curves at well logging resolution (typically 0.125 m), while seismic data refers to raw seismic data and extracted seismic attributes at seismic resolution (typically 2 ms). Scale coarsening is the process of converting high-resolution well data to match low-resolution seismic data. This method is a distance-weighted combined weighting method. Specifically, within a fixed-size sliding window, the lithofacies that appear most frequently are counted, and the lithofacies with the highest frequency (and thus the largest weight) within the window are assigned to the central sampling point. Based on this, the concept of distance weighting is introduced, with lithofacies closer to the center point having a higher weight. This is used to update the weights, ultimately coarsening the lithofacies data from the well logging to the scale of the seismic data while maintaining a smooth lithofacies transition.

[0083] In some executable embodiments, the seismic property volume includes an instantaneous amplitude property volume, an instantaneous frequency property volume, and a sweet spot property volume.

[0084] In some executable embodiments, step S3 includes:

[0085] The pseudo-well curves of different seismic attribute bodies at the oil well location are obtained to form a training set.

[0086] Furthermore, the well location includes the well point. Pseudo-well curves of different seismic attribute data volumes at the well point are extracted and used as input data for the training process, i.e., the training set.

[0087] A supervised learning algorithm and a training set are used to train a neural network model. The trained neural network model is then used to classify earthquake data to obtain lithofacies classification results. The lithofacies classification results include several lithofacies types. For each lithofacies type, an inverse distance weighted algorithm is used to calculate the lithofacies probability volume.

[0088] Specifically, a trained neural network model is obtained by using supervised learning algorithms such as kNN, and the most frequent label type among these k neurons will be used as the prediction result output.

[0089] When using the kNN algorithm, majority voting is generally used to determine the prediction result. If the number of labels is the same, a distance-weighted strategy is employed, weighting the votes based on the distance between each neighboring unit and the sample to be predicted. This avoids the result of a consistent number of labels. The label type is all lithofacies types on the lithofacies curve. The prediction result output is the predicted lithofacies type.

[0090] The trained neural network is used to classify the seismic data of the target layer in the entire work area, and the probabilities of different lithofacies predicted by these k neurons are calculated based on the inverse distance weighted algorithm.

[0091] The lithofacies probability volume represents the predicted probability of the same lithofacies at different locations within the work area, and its dimension is the same as that of the original seismic data and attribute volume. The lithofacies type is determined based on the actual lithology and fluid distribution within the work area. Taking common sandstone and mudstone strata as an example, lithofacies types can include oil-bearing sandstone facies, oil-water co-layer facies, water-bearing sandstone facies, sandstone-mudstone transitional facies, and mudstone facies.

[0092] In some executable embodiments, step S4 includes:

[0093] Based on the lithofacies classification results, a rock physics model is established, pseudo-well curves are obtained, and the target parameters of the initial seismic inversion model are determined.

[0094] Furthermore, rock physics modeling is carried out to obtain pseudo-well curves of physical properties such as pore structure and permeability, and to determine several target parameters for the initial modeling of the inversion.

[0095] In one executable embodiment, taking the direct inversion of rock matrix modulus, fluid modulus, and pore structure parameters from a sandstone-mudstone stratum as an example, the initial model to be constructed includes seven parameters: matrix bulk modulus, matrix shear modulus, fluid modulus, matrix density, fluid density, porosity, and equivalent pore aspect ratio. However, some parameters are coupled with each other. After rock physical modeling and analysis, four independent and uncoupled physical property parameters can be determined: clay content, water saturation, porosity, and equivalent pore aspect ratio. The initial model of the seven physical property parameters to be established can be directly calculated from these four parameters. Therefore, the initial model of the target physical property parameters established using the lithofacies probabilistic inverse mapping method includes four physical property parameters: clay content, water saturation, porosity, and equivalent pore aspect ratio.

[0096] Based on the lithofacies curves and target parameters, statistical rock physics analysis is performed on different lithofacies types to obtain the histogram distribution and probability density distribution of different physical properties in different lithofacies.

[0097] Furthermore, based on lithofacies curves and well physical property parameters, statistical rock physics analysis of different lithofacies is carried out to obtain histogram distributions and N-dimensional probability density distributions of different physical property parameters in different lithofacies types, where N is the number of physical property parameters involved in statistical rock physics.

[0098] Based on the histogram distribution and probability density distribution, a posterior probability density space is established.

[0099] Histogram distribution and probability density distribution are the foundation for establishing the posterior probability density space. Histograms help estimate the probability density function through lithofacies distribution characteristics, while the prior probability density, combined with the likelihood function (which describes the relationship between lithofacies distribution and physical property parameters), forms the posterior probability density through Bayes' theorem.

[0100] In some executable embodiments, the probability density distribution includes an N-dimensional probability density distribution.

[0101] The posterior probability density space includes an N-dimensional space of physical property parameters, covering the effective range of values ​​for N physical property parameters.

[0102] Where N is the number of physical property parameters involved in the statistical rock physics analysis.

[0103] In some executable embodiments, step S5 includes:

[0104] Based on the posterior probability density space, the posterior probabilities of different lithofacies corresponding to each point in the posterior probability density space are calculated and normalized according to the probability density distribution.

[0105] Specifically, prior probability values ​​are assigned to each point in the N-dimensional parameter space based on the histogram statistics of the physical property parameters. Based on the obtained N-dimensional parameter space, according to Bayes' theorem and probability density distribution, the posterior probability of different lithofacies corresponding to each parameter combination at each point in the space is calculated and normalized.

[0106] By iterating through the posterior probability of each sampling point in the seismic data, the lithofacies probability ratio of each sampling point in the seismic data is obtained.

[0107] Specifically, each sampling point is a point recorded in the seismic trace.

[0108] Based on the Monte Carlo algorithm and probability density function, M combinations of physical property parameters that conform to the preset parameter distribution are generated.

[0109] Based on the rejection-acceptance sampling algorithm and the lithofacies probability ratio of the current sampling point in the seismic data, m combinations of physical parameters that meet the conditions are selected from M combinations of physical parameters within a preset tolerance.

[0110] Specifically, the formula for combining M physical property parameters is as follows:

[0111]

[0112] in, Indicates based on probability density function Produce M samples M represents the total number of samples generated. This represents the probability density function used to generate samples.

[0113] The combination of m physical property parameters is used as the initial model for direct inversion of physical property parameters. This enables the allocation of different physical property parameters corresponding to different lithofacies.

[0114] In one executable embodiment, each sampling point within the seismic data includes every point recorded in the seismic trace, with a sampling rate of 2 ms.

[0115] In one executable embodiment, taking an oil-bearing sandstone-mudstone formation as an example, this embodiment selects formations with a clay content of less than 35% and a water saturation of less than 0.55 as oil-bearing sandstone facies, defines formations with a clay content of less than 35% and a water saturation between 0.55 and 0.7 as oil-water co-concrete facies, defines formations with a clay content of less than 35% and a water saturation of more than 0.7 as water-bearing sandstone facies, defines formations with a clay content between 35% and 60% as sandstone-mudstone transitional facies, and defines formations with a clay content of more than 60% as mudstone facies.

[0116] Figure 2 Based on the statistical petrophysical analysis results of the above facies classification, it is evident that a simple normal distribution model cannot adequately describe the probability density relationships between most physical property parameters. However, probability density estimation based on kernel functions can effectively describe such complex probabilistic models. In this model, each row represents a seismic facies, and each column represents a combination of physical property parameters. Based on this analysis, statistical petrophysical analysis of different facies can be conducted using lithofacies curves and physical property parameters, yielding histogram distributions and N-dimensional probability density distributions of different physical property parameters across different facies types. Figure 3 The probability density distribution corresponding to one of the attribute combinations (porosity and equivalent porosity aspect ratio) shows that there is a certain overlap between different phases, which increases the uncertainty of the process of identifying lithofacies by using physical property parameters, but it is difficult to have a negative impact on the inverse mapping process.

[0117] After completing the statistical rock physics analysis, seismic attributes such as instantaneous amplitude, instantaneous frequency, relative impedance, trace integral, and sweet spot are used in the training of the neural network. The lithofacies classification results identified by the trained neural network are as follows: Figure 4As shown (classification results for water-bearing sandstone, oil-bearing sandstone, and mudstone were selected). Based on the lithofacies classification results and the corresponding probability volumes, the inverse mapping of physical property parameters was achieved using steps S4-S5. The initial models for porosity and pore structure, two physical property parameters obtained from the inverse mapping of lithofacies probabilities, are as follows: Figure 5 and Figure 6 As shown, this is the final high-precision low-frequency model.

[0118] Due to the limited number of wells in the work area (logging data from 4 wells), it is difficult to obtain the plane variation function using direct logging data. Therefore, hastily establishing an initial model of physical property parameters using geostatistical methods does not have high reliability. Thus, a conventional initial model of physical property parameters is established using trend analysis in the vertical direction and discrete smoothing interpolation in the plane. The results are as follows: Figure 7 As shown, a comparison with the actual physical property parameters of a well at the well point reveals a significant error. The initial physical property parameter model established using the inverse mapping of lithofacies probabilistics is as follows: Figure 8 As shown, the results are in good agreement with the calculated physical property parameters at the well point, proving the effectiveness of the method.

[0119] The initial model established by this invention based on inverse probabilistic mapping of lithofacies not only avoids the serious problems of traditional inter-well interpolation models, but also does not require reliance on accurate variability functions to simulate spatial geological variations. Furthermore, the vertical variation of physical property parameters is not as pronounced as that of elastic parameters, and their spatial variation is more easily controlled by geological units such as lithofacies and sedimentary facies. Therefore, the initial physical property model obtained by using vertical trends and planar discretization smoothing interpolation based on a stratigraphic sequence framework exhibits significant errors at well points, resulting in low reliability of the initial model established by traditional methods. In contrast, the initial physical property model established using inverse probabilistic mapping of lithofacies shows better consistency with the well point model, and the spatial variation of the parameters conforms to the lithofacies distribution, making it more geologically significant.

[0120] The present invention also provides an apparatus for establishing an initial seismic inversion model based on inverse lithofacies probability mapping. The apparatus includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of the method for establishing an initial seismic inversion model based on inverse lithofacies probability mapping as described above.

[0121] The above embodiments only illustrate preferred embodiments of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can freely combine the above technical features without departing from the concept of the present invention, and can also make several modifications and improvements, all of which fall within the protection scope of the present invention. Therefore, all equivalent transformations and modifications made with respect to the scope of the claims of the present invention should fall within the scope of the claims of the present invention.

Claims

1. A method for establishing an initial model of seismic inversion based on inverse mapping of lithofacies probability, characterized in that, The method comprises the following steps: S1, collecting and processing oil test data, seismic data, well logging data and horizon data to obtain post-stack seismic data and lithofacies curve; S2, obtaining a plurality of seismic attribute volumes from the post-stack seismic data, and obtaining a pseudo-well curve of each seismic attribute volume in combination with the lithofacies curve; S3, training a neural network model using the pseudo-well curve, and outputting a lithofacies classification result and a lithofacies probability volume by using the trained neural network model in combination with the seismic data; S4, analyzing the lithofacies classification result and the lithofacies probability volume to obtain a probability density distribution of different physical parameters, and establishing a posterior probability density space of the physical parameters; S5, based on the posterior probability density space and the probability density distribution, traversing the seismic data, and using a Monte Carlo algorithm and a rejection-acceptance sampling to generate a qualified physical parameter combination to obtain the initial model of the seismic inversion.

2. The lithofacies probability inverse mapping based seismic inversion initial model building method according to claim 1, characterized in that, The step S1 comprises: collecting the oil test data, the seismic data, the well logging data and the horizon data in the work area; preprocessing the seismic data, the well logging data and the horizon data to obtain a time-depth relationship after well-seismic; obtaining a physical parameter distribution through the time-depth relationship; dividing the lithofacies of the oil well according to the physical parameter distribution to obtain a lithofacies curve of the well; performing migration stacking on the seismic data to obtain the post-stack seismic data.

3. The method according to claim 1, wherein, The step S2 comprises: obtaining a seismic attribute volume based on the post-stack seismic data according to a preset condition; matching the resolution of the lithofacies curve with the inherent resolution of the post-stack seismic data by using a scale coarsening method to obtain the pseudo-well curve.

4. The method according to claim 3, wherein, The seismic attribute volume comprises an instantaneous amplitude attribute volume, an instantaneous frequency attribute volume and a sweet spot attribute volume.

5. The lithofacies probability inverse mapping based seismic inversion initial model building method according to claim 1, characterized in that, The step S3 comprises: obtaining the pseudo-well curve of different seismic attribute volumes at the well site of the oil well to obtain a training set; training the neural network model using a supervised learning algorithm and the training set; classifying the seismic data using the trained neural network model to obtain the lithofacies classification result; the lithofacies classification result comprises a plurality of lithofacies classifications; each lithofacies classification is calculated in combination with an inverse distance weighted algorithm to obtain the lithofacies probability volume.

6. The lithofacies probability inverse mapping based seismic inversion initial model building method according to claim 1, characterized in that, The step S4 comprises: establishing a rock physics modeling according to the lithofacies classification result, obtaining a well curve, and determining a target parameter of the initial model of the seismic inversion; statistically analyzing the rock physics of different lithofacies according to the lithofacies curve and the target parameter to obtain a histogram distribution and a probability density distribution of different physical parameters in different lithofacies; establishing the posterior probability density space according to the histogram distribution and the probability density distribution.

7. The method according to claim 6, wherein, The probability density distribution comprises an N-dimensional probability density distribution; The posterior probability density space comprises an N-dimensional physical parameter space covering an effective value range of N physical parameters; wherein N is the number of physical parameters participating in the statistical rock physics analysis.

8. The method according to claim 6, wherein, The step S5 comprises: Based on the posterior probability density space, a posterior probability of different lithofacies of the combination of the petrophysical parameters corresponding to each point in the posterior probability density space is calculated according to the probability density distribution and normalized; The posterior probability of each sampling point in the seismic data is traversed to obtain a lithofacies probability ratio of each sampling point in the seismic data; M combinations of petrophysical parameters conforming to a preset parameter distribution are generated according to a Monte Carlo algorithm and a probability density function; m combinations of petrophysical parameters meeting a condition are selected from the M combinations of petrophysical parameters within a preset tolerance according to a rejection-acceptance sampling algorithm and the lithofacies probability ratio of the current sampling point in the seismic data; The m combinations of petrophysical parameters are used as initial models of direct inversion of petrophysical parameters.

9. The lithofacies probability inverse mapping based seismic inversion initial model building method according to claim 8, characterized in that, Each sampling point in the seismic data includes each point of a seismic trace record, and a sampling rate is 2 ms.

10. An apparatus for establishing an initial model for seismic inversion based on inverse mapping of lithofacies probability, the apparatus comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the steps of the method for establishing an initial model of seismic inversion based on lithofacies probability inverse mapping according to any one of claims 1-9 when executing the program.

Images