Method and device for establishing three-dimensional sandstone probability body

Abstract

The invention provides a method and a device for establishing a three-dimensional sandstone probability body. The method comprises the following steps: integrating a well-seismic database; constructing a three-dimensional construction framework model; converting the time domain seismic data volume into a depth domain seismic data volume; extracting a target seismic attribute according to a preset rule; sampling a target seismic attribute to a three-dimensional construction framework model; performing multi-attribute clustering fusion on the seismic attribute values of the three-dimensional construction framework model by using a preset neural network model, and establishing a sedimentary unit model; calculating a variation function between the sandstone content data for each partition of the sedimentary unit model; and performing interpolation by using a Kriging algorithm, and predicting the sandstone content of an unknown point to obtain a three-dimensional sandstone probability body model. According to the method, a multi-type and multi-attribute fused three-dimensional sandstone probability body is established, the value type of reservoir facies and the value of reservoir parameters are constrained by endowing each grid with a probability value, and reliable trend constraint conditions are provided for geologic modeling of low-well-control and strong-heterogeneous reservoirs.

Description

Technical Field

[0001] The invention belongs to the field of sedimentary reservoir evaluation in petroleum geological exploration, and in particular relates to a method and a device for establishing a three-dimensional sandstone probability body. Background Art

[0002] In recent years, relying on computer technology and in-depth research in mathematical geology, 3D geological modeling technology has developed rapidly. 3D geological modeling technology is a digital model that uses geostatistics, spatial analysis and prediction methods, based on the research results of seismic, well logging, geology, dynamics, etc., and uses computer technology to characterize the spatial distribution law and change characteristics of geological parameters.

[0003] Commonly used modeling algorithms include deterministic algorithms and random algorithms. Target-based and pixel-based random algorithms have been widely used in this field, providing basic conditions for three-dimensional reservoir characterization. The pixel-based sequential indicator modeling method mainly uses the variation function between two points to reflect the heterogeneity of the reservoir structure. This method is suitable for dense well network areas in the oil and gas field development stage. For areas with few wells and strong heterogeneity in the exploration and evaluation stage, the global variation function is difficult to characterize the changes in the river channel. The target-based modeling method can add prior geological knowledge as a condition to the simulation to reproduce the knowledge of geologists to the greatest extent. This method has been widely used.

[0004] At present, the phase boundary constraint and single trend constraint are mainly used to constrain the simulation process in reservoir modeling, which improves the rationality of the modeling results to a certain extent. However, for areas with little drilling data, the internal variation function of the boundary is not properly controlled and the single trend is not fully considered. This method cannot depict the reasonable spatial distribution law of the reservoir. In addition, the multi-condition constraint modeling method of weighted average conversion of multiple trends has also been widely used, but this method is difficult to determine the weight of each attribute on each grid, and the multi-trend integration process is difficult. Summary of the invention

[0005] In order to solve the problems encountered by traditional modeling constraint methods under the conditions of few wells and strong heterogeneity, the present invention proposes a method and device for establishing a three-dimensional sandstone probability body.

[0006] A method for establishing a three-dimensional sandstone probability volume comprises the following steps:

[0007] An integrated well-seismic database, wherein the data of the well-seismic database includes at least well information of the study area, geological research results data, seismic results data, and well logging interpretation results data, wherein the seismic results data includes at least two depth domain interpretation fault data, two depth domain interpretation layer data, two time domain interpretation layer data, and one time domain seismic data volume;

[0008] Using the earthquake results data to construct a three-dimensional structural framework model;

[0009] Converting the time-domain seismic data volume into a depth-domain seismic data volume, and updating the well-seismic database;

[0010] Extracting target seismic attributes from the updated well-seismic database according to preset rules, wherein the target seismic attributes can reflect reservoir changes;

[0011] Sampling the target seismic attribute to the three-dimensional structural framework model, and assigning seismic attribute values ​​to the three-dimensional structural framework model;

[0012] The preset neural network model is used to perform multi-attribute clustering and fusion on the seismic attribute values ​​of the three-dimensional structural framework model to establish a sedimentary unit model;

[0013] The variogram between the sandstone content data is calculated for each partition of the sedimentary unit model, and the sandstone content data of the unsampled position of the sedimentary unit model is estimated;

[0014] According to the known sandstone content distribution data in the variogram and logging interpretation data, the Kriging algorithm is used to interpolate on the three-dimensional structural framework grid to predict the sandstone content of unknown points and obtain a three-dimensional sandstone probability volume model.

[0015] Furthermore, a three-dimensional structural framework model is constructed using the seismic data, including:

[0016] The three-dimensional structural framework model is constructed by the two depth domain interpretation levels and the two depth domain interpretation faults.

[0017] Furthermore, converting the time domain seismic data volume into depth domain seismic volume data includes:

[0018] The time domain seismic data volume is converted into a depth domain seismic data volume using a preset velocity model, wherein the preset velocity model is:

[0019]

[0020] Among them, V1 and V2 represent two time domain interpretation level data respectively, Z represents depth, Z1 and Z2 are two depth domain interpretation level data, and the v interp It is the depth domain seismic data volume converted from the time domain seismic data volume.

[0021] Furthermore, in the updated well-seismic database, target seismic attributes are extracted according to preset rules, including:

[0022] Extract data on seismic attributes related to lithofacies and porosity and permeability;

[0023] After normalizing the data of the seismic attributes and unifying the dimensions, a spatial intersection diagram of the seismic attributes, porosity, permeability, and lithofacies is drawn;

[0024] Screening out seismic attributes for distinguishing sandstone facies and mudstone facies according to the spatial intersection diagram;

[0025] Among the screened seismic attributes, the attributes with the highest similarity values ​​between the spatial variation trend and the sand body distribution map and oil-bearing area map of the geological research results data are selected as the target seismic attributes.

[0026] Furthermore, the preset neural network model is used to perform multi-attribute clustering fusion on the seismic attribute values ​​of the three-dimensional structural framework model to establish a sedimentary unit model, including:

[0027] The seismic attribute values ​​of the three-dimensional structural framework model are input into the preset neural network model to perform multi-attribute clustering fusion;

[0028] Adjust the parameters of the preset neural network model until the loss function converges to a preset level or reaches a predetermined number of training iterations;

[0029] The output sedimentary unit model is compared with the well information and geological research results data. When the conformity between the output sedimentary unit model and the well information and geological research results data reaches a preset value, the adjustment is stopped to obtain a determined sedimentary unit model.

[0030] Furthermore, the preset neural network model includes an encoder and a decoder, wherein the encoder is represented as:

[0031] h1=Relu(W1x+b1)

[0032] z=W2h1+b2

[0033] Where x is the input data, z is the low-dimensional mapping of the input data, W1 is the first-layer weight matrix of the encoder, b1 is the first bias, W2 is the second-layer weight matrix of the encoder, and b2 is the second bias;

[0034] The loss function of the neural network model is:

[0035]

[0036] Where N is the number of samples, xi is the original input of the i-th sample, and xi′ is the corresponding reconstructed output.

[0037] Furthermore, the calculation of the variogram between the sandstone content data for each partition of the sedimentary unit model respectively, and the estimation of the sandstone content data of the unsampled position of the sedimentary unit model, include:

[0038] Collect sandstone content data for each subarea of ​​the sedimentary unit model;

[0039] Calculate the semivariance between each pair of sandstone content data;

[0040] Group the semivariance values ​​according to the distance bandwidth;

[0041] Calculate the mean or variation of the semivariance value within each distance bandwidth to form different lag values ​​of the variogram;

[0042] Fitting the model to the variogram;

[0043] The fitted variogram is used for spatial interpolation to estimate the sandstone content data at unsampled locations in the sedimentary unit model.

[0044] Furthermore, the method also includes visualizing the obtained three-dimensional sandstone probability volume model.

[0045] On the other hand, the present invention also discloses a device for establishing a three-dimensional sandstone probability body, including an integrated well-seismic database module, a three-dimensional structural framework model construction module, a data body conversion module, a target seismic attribute extraction module, a sampling module, a sedimentary unit model establishment module, a variation function module, and a three-dimensional sandstone probability body model generation module.

[0046] Based on the above technical solution, the beneficial effects of the present invention compared with the prior art are as follows:

[0047] In order to solve the problems encountered by traditional modeling constraint methods under the conditions of few wells and strong heterogeneity, the embodiment of the present invention first establishes a well-seismic database, extracts seismic attributes such as frequency, amplitude, phase, and wave impedance, and performs geological interpretation on the seismic attributes. Through the seismic response characteristics of the reservoir, the seismic attributes that can reflect reservoir changes are preferred; then, unsupervised artificial neural network technology is used to perform multi-attribute clustering fusion, and a sedimentary unit model is established in combination with prior geological knowledge results; finally, a three-dimensional sandstone probability body is established using geostatistical technology based on the Kriging algorithm. The present invention establishes a three-dimensional sandstone probability body with multiple types and multiple attributes, and constrains the value type of the reservoir phase and the value size of the reservoir parameter by assigning a probability value to each grid. The present invention can provide reliable trend constraints for geological modeling of low well control and strong heterogeneity reservoirs, and more accurately and reliably reproduces the actual reservoir geological characteristics compared to traditional methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 This is a schematic flow chart of a method for establishing a three-dimensional sandstone probability volume in the first embodiment of the present invention;

[0050] Figure 2 This is a flow chart of step S140 in the first embodiment of the present invention;

[0051] Figure 3 This is a schematic diagram of the correlation between the seismic attributes of the envelope and the porosity in the first embodiment of the present invention;

[0052] Figure 4 This is a flow chart of step S160 in the first embodiment of the present invention;

[0053] Figure 5 This is a schematic diagram of the output deposition unit model in the first embodiment of the present invention;

[0054] Figure 6 Schematic diagram of a three-dimensional sandstone probabilistic volume model in Embodiment 1 of the present invention;

[0055] Figure 7 It is a schematic diagram of a lithofacies model established by using the constraints of the three-dimensional sandstone probabilistic volume model of the present invention in the first embodiment of the present invention;

[0056] Figure 8 This is a schematic diagram of a porosity model established by using the constraints of the three-dimensional sandstone probabilistic volume model of the present invention in the first embodiment of the present invention;

[0057] Fig. 9 It is a schematic diagram of a permeability model established by using the three-dimensional sandstone probabilistic volume model constraint of the present invention in the first embodiment of the present invention;

[0058] Fig.10 This is a schematic diagram of experimental results of applying the three-dimensional sandstone probabilistic volume model of the present invention to modeling a reservoir in a certain oil field in the first embodiment of the present invention;

[0059] Fig.11 This is a schematic structural diagram of a device for establishing a three-dimensional sandstone probability volume in the second embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0061] A method for establishing a three-dimensional sandstone probability volume, such as Figure 1 As shown, it includes steps 110 to 180:

[0062] Step 110, integrating the well-seismic database, wherein the data in the well-seismic database at least includes well information of the study area, geological research results data, seismic results data, and well logging interpretation results data.

[0063] Among them, well information at least includes well location and well trajectory; geological research results data at least include sand body plane isodistribution map (abbreviated as: sand body distribution map) and oil-bearing area distribution map (abbreviated as: oil-bearing area map); seismic-related results data at least include two depth domain interpretation fault data, two depth domain interpretation layer data, two time domain interpretation layer data, and one time domain seismic data body; well logging interpretation results data at least include comprehensive interpretation well logging curves (porosity, permeability, etc.), single well lithofacies (sandstone, mudstone) data, etc.

[0064] Step 120, constructing a three-dimensional structural framework model using the seismic data.

[0065] In the subsequent steps, in order to analyze multiple types of data such as well information, seismic data, and logging interpretation data together, a three-dimensional structural framework grid model needs to be established to accommodate multiple types of data. In practical applications, the three-dimensional structural framework model is constructed by the two depth domain interpretation levels and two depth domain interpretation faults, for example, the top and bottom interfaces and faults of a certain section of an oil layer in a certain place are presented in three-dimensional form.

[0066] Step 130: convert the time domain seismic data volume into a depth domain seismic data volume, and update the well seismic database.

[0067] Since the propagation of seismic waves in the strata is inevitably affected by the properties of the strata (lithology, physical properties, fluid properties, etc.), the difference in the properties of the underground reservoir will lead to differences in the wave impedance of the reservoir, which in turn causes seismic response. Seismic attributes such as frequency, amplitude, phase, and wave impedance may reflect geological characteristics such as lithology, lithofacies, rock continuity, and porosity. Based on this principle, it is necessary to analyze the relationship between seismic attributes and lithofacies, porosity, and permeability, and combine the geological research results data of the study area to select multiple seismic attributes that can reflect the properties of the reservoir. However, in actual applications, the time domain seismic data body is time domain data, and the well information and geological research results data are all depth domains, so it is necessary to perform time-depth conversion on the time domain seismic data body and convert it into a depth domain seismic data body, and then extract the depth domain seismic attributes, which is convenient for analyzing the relationship between the depth domain seismic attributes and lithofacies, porosity, and permeability.

[0068] In order to realize the conversion of time coordinates in seismic data into depth coordinates, the inventor established a velocity model using two time domain interpretation layer data and two depth domain interpretation layer data to calculate the velocities of the two layers respectively, and filled the velocity changes between the two layers through linear interpolation method.

[0069] The default speed model is:

[0070]

[0071] Among them, V1 and V2 represent two time domain interpretation level data respectively, Z represents depth, Z1 and Z2 are two depth domain interpretation level data, and the v interp It is the depth domain seismic data volume converted from the time domain seismic data volume.

[0072] Step 140: extract target seismic attributes from the updated well-seismic database according to preset rules, wherein the target seismic attributes can reflect reservoir changes.

[0073] Combination Figure 2 As shown, this step includes sub-steps S141-S144:

[0074] Step S141, extracting data of seismic attributes related to lithofacies and porosity and permeability.

[0075] In some embodiments, the seismic attributes related to facies and porosity and permeability are sweet spots, instantaneous frequency, root mean square amplitude, envelope, inversion, etc.

[0076] Step S142, after normalizing the seismic attribute data and unifying the dimensions, a spatial intersection diagram of seismic attributes, porosity, permeability, and lithofacies is drawn.

[0077] Step S143: Screening out seismic attributes for distinguishing sandstone facies and mudstone facies according to the spatial intersection diagram.

[0078] In some embodiments, Figure 3 As shown in the figure, taking the correlation between the seismic attributes of the envelope and porosity as an example, the value range of the seismic attributes of the envelope corresponding to sandstone is 0.3-0.6, and the value range of the seismic attributes of the envelope corresponding to mudstone is 0.18-0.22 and 0.63-0.86. Therefore, when the seismic attribute value of the envelope is 0.3-0.6, it may correspond to the sandstone phase, and different value intervals can be color-coded to facilitate comparison with geological research results maps.

[0079] Step S144, selecting the attribute with the highest similarity value between the spatial variation trend and the sand body distribution map and the oil-bearing area map of the geological research results data from the screened seismic attributes as the target seismic attribute.

[0080] In practical applications, similarity calculation methods can be used to compare the characteristics of seismic attributes and geological data. Example calculation steps:

[0081] Step S1441: For each seismic attribute, the similarity value between it and the sand body distribution map is calculated using the Pearson correlation coefficient (r).

[0082]

[0083] Among them, Xi is the seismic attribute, Yi is the sand body content

[0084] Step S1442: Similarly, for each seismic attribute, the similarity value between it and the oil-bearing area map is calculated using the same similarity calculation method.

[0085] Step S1443, using weighted average to combine the similarity values ​​calculated in step S1441 and step S1442 to obtain a comprehensive similarity score.

[0086] The comprehensive similarity scores of all seismic attributes are compared, and the seismic attribute with the highest similarity score is selected as the target seismic attribute. This attribute is most similar to the sand body distribution map and the oil-bearing area map in terms of spatial variation trend.

[0087] In some embodiments, the target seismic attributes obtained are sweet spot, root mean square amplitude, inversion, instantaneous frequency.

[0088] Step 150: sampling the target seismic attributes to the three-dimensional structural framework model, and assigning seismic attribute values ​​to the three-dimensional structural framework model.

[0089] In some embodiments, the sweet spot, root mean square amplitude, inversion, and instantaneous frequency seismic attribute data bodies are extracted into a three-dimensional structural framework grid model, that is, the seismic attribute data bodies are sampled into the three-dimensional structural framework grid, and seismic attribute values ​​are assigned to the three-dimensional structural framework grid. The seismic attributes are assigned to the three-dimensional structural framework grid, which can well reflect the changes in lithofacies in space.

[0090] Step 160, using a preset neural network model to perform multi-attribute clustering and fusion on the seismic attribute values ​​of the three-dimensional structural framework model to establish a sedimentary unit model.

[0091] like Figure 4 As shown, this step specifically includes sub-steps S161-S163, specifically:

[0092] Step S161, inputting the seismic attribute values ​​of the three-dimensional structural framework model into a preset neural network model to perform multi-attribute clustering fusion.

[0093] This neural network model has some advantages in multi-attribute clustering fusion:

[0094] 1. Automatic feature extraction: Autoencoders can automatically extract features by learning the intrinsic representation of data. This is very useful for multi-attribute clustering fusion because it helps the system identify relevant features in the data for better clustering.

[0095] 2. Dimensionality reduction and reconstruction: The encoder part maps high-dimensional data to a low-dimensional space (z) to reduce the dimension of the data. In the reconstruction phase, the decoder maps the low-dimensional representation back to high-dimensional data, thereby retaining the important information of the data. This dimensionality reduction and reconstruction process can reduce redundant information and help improve clustering results.

[0096] 3. Nonlinear modeling: The fully connected layers in the encoder and decoder can learn nonlinear relationships, which is very useful for dealing with complex data distributions and nonlinear relationships between features.

[0097] The preset neural network model includes an encoder and a decoder, where the encoder is composed of a fully connected layer and can be expressed as:

[0098] h1=Relu(W1x+b1)

[0099] z=W2h1+b2

[0100] Among them, x is the input data, z is the low-dimensional mapping of the input data, W1 is the first-layer weight matrix of the encoder, b1 is the first bias, W2 is the second-layer weight matrix of the encoder, and b2 is the second bias.

[0101] These parameters are the weight matrix and bias term of the neural network, which are learned by training the neural network model. During the training process, the model adjusts these parameters based on the input data (seismic attribute values) and the target data (data that is the same as the input data, used for reconstruction) so that the model can best reconstruct the input data.

[0102] The structure of the decoder is opposite to that of the encoder, and a fully connected layer can be used. The decoder decodes the low-dimensional representation z into the reconstructed feature x'. Assuming that the first-layer weight matrix of the decoder is W3, the bias is b3, and the second-layer weight matrix is ​​W4, the bias is b4, then the output of the decoder can be expressed as:

[0103] h2=Relu(W3z+b3)

[0104] x'=W4h2+b4

[0105] The loss function of this neural network model is as follows to measure the error between the reconstructed output and the original input:

[0106]

[0107] Where N is the number of samples, xi is the original input of the i-th sample, and xi′ is the corresponding reconstructed output.

[0108] Step S162, adjusting the parameters of the preset neural network model until the loss function converges to a preset level or reaches a predetermined number of training iterations.

[0109] Adjusting neural network parameters is done by training the neural network model. The main goal of neural network training is to minimize the loss function, which measures the gap between the output of the model and the actual target, that is, to minimize the difference between the input data and the reconstructed data.

[0110] Initialization parameters: First, you need to initialize the parameters of the neural network, including the weight matrix (W1, W2) and the bias term (b1, b2). Usually, the parameters are initialized to small random values ​​to ensure that the network has enough randomness to start training.

[0111] Forward propagation: Forward propagation is performed on each training sample. This means passing the input data (seismic attribute value x) into the encoder part (fully connected layer) to obtain the encoded representation z. Then, z is passed to the decoder part to reconstruct the data. This will generate the output of the model, which is the reconstructed data.

[0112] Calculate loss: Use a loss function to calculate the difference between the model output and the actual target data. In this case, the loss function is usually the mean squared error between the input data and the reconstructed data.

[0113] Backpropagation: Using the calculated loss, backpropagation is performed to calculate the gradients of the parameters to minimize the loss function.

[0114] Parameter update: Use gradient descent to update the parameters (W1, b1, W2, b2). Typically, multiple forward propagation, loss calculation, and backpropagation are performed on each training sample, and then the parameters are updated according to the gradient descent rule.

[0115] Iterative training: Repeat the forward propagation, loss calculation, backpropagation, and parameter update steps until the loss converges to a satisfactory level or reaches a predetermined number of training iterations.

[0116] Step S163, comparing the difference between the output sedimentary unit model and the well information and geological research results data. When the conformity between the output sedimentary unit model and the well information and geological research results data reaches a preset value, stop adjusting and obtain a determined sedimentary unit model.

[0117] In some embodiments, the seismic attribute data of the sweet spot, root mean square amplitude, inversion, and instantaneous frequency determined in step S150 are input into the preset neural network model in step S161, the parameters are adjusted, and the output sedimentary unit model is compared with the well information and geological research results data until the consistency between the sedimentary unit model and the well information and geological research results reaches a preset value, and the training is stopped.

[0118] The distribution of the inventor's sedimentary unit model in low-dimensional space is highly similar to the geological research results data, with a similarity of >80%, indicating that the output reflects the geological understanding results to a certain extent. Therefore, in this embodiment, after 1000 iterations of training, the error is limited to 10%, and when the preset value is set to 80%, the output is divided into 5 types of sedimentary unit models. Figure 5 As shown, in the output sedimentary unit model, the geological cognition results corresponding to the Class I area are areas with high sandstone development and high test oil production, the geological cognition results corresponding to the Class II area are areas with lower sandstone development and lower test oil production than the Class I area, and the geological cognition results corresponding to the Class III area, the Class IV area, and the Class V area are areas with more mudstone phase changes, and the mudstone phase gradually becomes lower in these three types of areas. The sedimentary unit model of the present invention integrates the microfacies characteristics of multiple seismic attributes and highlights the spatial distribution of sedimentary phases.

[0119] Step 170 , respectively calculating the variogram between the sandstone content data for each partition of the sedimentary unit model, and estimating the sandstone content data of the unsampled position of the sedimentary unit model.

[0120] The variogram between sandstone content data is calculated for each partition of the sedimentary unit model. This function is used to describe how the variability between data points changes with increasing distance. It is usually used for spatial interpolation of geological data to understand the spatial variability of data.

[0121] It is necessary to collect sandstone content data, assign each data point to a three-dimensional spatial model, and calculate the distance between each pair of data points. If the spatial direction between the data points has an impact on the variability, you can consider calculating the direction angle. For each pair of data points, calculate the semi-variance between them. The semi-variance represents the spatial correlation or difference between the data points. Group the semi-variance values ​​according to the distance bandwidth. The distance bandwidth is a custom parameter, indicating that the semi-variance values ​​within different distance ranges should be divided into different groups. Calculate the mean or variation of the semi-variance values ​​within each distance bandwidth. These values ​​constitute the different lags of the variogram. Fit the variogram model, and use the semi-variance function model (which can be an exponential model, a Gaussian model, a spherical model, etc.) to describe the change of the semi-variance value with distance. Model fitting can show the characteristics of spatial variability, such as the degree of variation, range, etc. Use the fitted variogram to perform spatial interpolation to estimate the sandstone content data at the unsampled locations of the sedimentary unit model.

[0122] Step 180, based on the known sandstone content distribution data in the variogram and the logging interpretation data, the Kriging algorithm is used to interpolate on the three-dimensional structural framework grid to predict the sandstone content of the unknown point and obtain a three-dimensional sandstone probability volume model.

[0123] Using the Kriging algorithm, we interpolate the known sandstone content distribution data in the variogram and the logging interpretation data on the 3D structural framework grid to predict the sandstone content of the unknown points. For each grid cell, we calculate the probability distribution of the Kriging prediction values ​​of the surrounding points. We visualize the established 3D sandstone probability volume model, use color coding to represent different probability areas, and intuitively understand the uncertainty of the sandstone content prediction of the unknown points.

[0124] For each unknown point of sandstone distribution within each 3D model grid cell, the Kriging interpolation method is used for prediction. Kriging is a geostatistical method used to estimate the value of an unknown location, taking into account the weight and distance of known points. In this step, for each unknown point, the Kriging predicted values ​​of its surrounding known points are used to estimate the sandstone content.

[0125] 1. Calculate the Kriging Estimate:

[0126] For each unknown point P_i, calculate the distance d_ij between it and the surrounding known points P_j.

[0127] Use the observed value Z_j of the known point P_j and the semivariogram function γ(d_ij) to calculate the Kriging estimate Z_i of the unknown point P_i:

[0128] Z_i=∑[λ_j*Z_j], where j is all known points and λ_j is the weight coefficient.

[0129] The weight coefficient λ_j can be determined by an optimization method (such as the least squares method) to minimize the mean square error between the Kriging estimate and the known point observations.

[0130] 2. Calculate the covariance:

[0131] Use the semivariogram function γ(d_ij) to calculate the covariance C_ij between the unknown point P_i and the surrounding known points P_j:

[0132] C_ij=Cov(Z_i,Z_j)=σ^2-(d_ij), where σ^2 is the global variance.

[0133] 3. Calculate the variance:

[0134] The variance Var_i of the unknown point P_i is calculated by the following formula:

[0135] Var_i=σ^2-∑[λ_j*C_ij], where j is all known points and λ_j is the weight coefficient.

[0136] 4. Calculate the probability distribution:

[0137] Using the Kriging estimates Zi and variance Var, we can construct the probability distribution of sandstone content. Assume that the sandstone content conforms to the normal distribution:

[0138] Mean μ_i=Z_i

[0139] Varianceσ^2_i=Var_i

[0140] The normal distribution of sandstone content at the unknown point P_i is obtained, where the mean is the Kriging estimate and the variance is the estimated variance.

[0141] 5. Visualize probability distribution:

[0142] Using the obtained mean and variance, use visualization tools to display the probability distribution. A common way is to use color maps or contour maps to represent different probability areas, where the density of colors or lines indicates the high and low probabilities.

[0143] a. Choose a visualization tool or software, such as a scientific computing tool (such as Python's Matplotlib library).

[0144] b. Create a color code for each unknown point, where the color is based on the probability distribution of sandstone content at that point. Typically, you can use cool colors to represent low probabilities and warm colors to represent high probabilities.

[0145] c. Plot these color-coded points or areas on the 3D construction framework grid to represent areas of different probabilities. This information can be presented using contour lines, gradient color maps, or stereograms to more clearly understand the distribution of uncertainty.

[0146] Finally, analyze the visualization results to understand how the probability distribution of sandstone content varies in space. Determine which areas have a high confidence in the sandstone content estimate and which areas may have greater uncertainty.

[0147] The established 3D sandstone probability volume integrates multiple types of seismic attributes and lithofacies data, and the probability value of each grid is 0 to 1. The 3D sandstone probability volume can be used as a trend constraint body in lithofacies and porosity and permeability modeling to improve the accuracy of reservoir modeling. Figure 6 The three-dimensional sandstone probability body model obtained by the embodiment of the present invention is illustrated. The area with probability greater than 0.5 represents the high sandstone area, and the area with probability less than 0.5 represents the low sandstone phase and mudstone phase. The established three-dimensional probability body model can constrain the value type of the lithofacies and the value size of the porosity and permeability.

[0148] In order to solve the problems encountered by traditional modeling constraint methods under the conditions of few wells and strong heterogeneity, the embodiment of the present invention first establishes a well-seismic database, extracts seismic attributes such as frequency, amplitude, phase, and wave impedance, and performs geological interpretation on the seismic attributes. Through the seismic response characteristics of the reservoir, the seismic attributes that can reflect reservoir changes are preferred; then, unsupervised artificial neural network technology is used to perform multi-attribute clustering fusion, and a sedimentary unit model is established in combination with prior geological knowledge results; finally, a three-dimensional sandstone probability body is established using geostatistical technology based on the Kriging algorithm. The present invention establishes a three-dimensional sandstone probability body with multiple types and multiple attributes, and constrains the value type of the reservoir phase and the value size of the reservoir parameter by assigning a probability value to each grid. The present invention can provide reliable trend constraints for geological modeling of low well control and strong heterogeneity reservoirs, and more accurately and reliably reproduces the actual reservoir geological characteristics compared to traditional methods.

[0149] Specifically, Figure 7-9 As shown, the sequential simulation algorithm is selected, and the sandstone probability body constraint is applied to establish the lithofacies model, porosity model, and permeability model. The trends of the lithofacies model and the porosity and permeability models are well consistent with the three-dimensional sandstone probability body, reflecting the effectiveness of the three-dimensional sandstone probability constraint body in controlling the model. It can be seen that the three-dimensional sandstone probability body obtained by using the embodiment of the present invention can be directly applied to lithofacies and porosity and permeability modeling as a trend constraint body, which takes into account the uncertainty of reservoir prediction and significantly improves the accuracy of reservoir modeling.

[0150] The inventors also conducted experiments using the embodiments of the present invention in modeling a reservoir in a certain oil field. The experimental results are as follows: Fig.10 As shown in Table 1, three sweet spots are divided according to the reservoir parameter prediction results. In the Class I area, three production wells are selected, the oil layer is well developed, and the actual drilling results are 100% consistent with the model; in the Class II area, one evaluation well is selected, the reservoir is well developed, the oil layer is poorly developed, and the actual drilling results are 90% consistent with the model. Compared with the traditional method, the geological model constructed by the three-dimensional probability body constraint method of the present invention reproduces the actual reservoir geological characteristics more accurately and reliably. The present invention can be applied and promoted in the reservoir evaluation stage of oil and gas exploration, and has reference significance for the three-dimensional geological modeling of other similar mineral reservoirs.

[0151] Table 1

[0152]

[0153]

[0154] In summary, the sandstone probability body is used in three-dimensional reservoir modeling to constrain the reservoir phase and reservoir parameter modeling. The three-dimensional sandstone probability body can be used as a three-dimensional trend. The probability of reservoir phase and reservoir parameter appearing in the grid is directly controlled by the sequential simulation method with trend and the sequential Gaussian indicator algorithm, respectively. It can be applied to lithofacies and porosity and permeability modeling, reducing the randomness and uncertainty of reservoir prediction, improving the reliability of model prediction, and making the final model meet the requirements of probability body trend distribution. The invention is suitable for oil and gas blocks in the exploration and evaluation stage with little drilling data, mainly seismic data, strong reservoir heterogeneity, and has reference significance for other mineral reservoir geological modeling.

[0155] Embodiment 2

[0156] In other embodiments, the present invention further discloses a device for establishing a three-dimensional sandstone probability volume model, comprising an integrated well-seismic database module 10, a three-dimensional structural framework model building module 20, a data volume conversion module 30, a target seismic attribute extraction module 40, a sampling module 50, a sedimentary unit model establishment module 60, a variogram module 70, and a three-dimensional sandstone probability volume model generation module 80, wherein:

[0157] The integrated well-seismic database module 10 is used to integrate the well-seismic database, the data of which include at least the well information of the study area, geological research results data, seismic results data, and well logging interpretation results data, wherein the seismic results data include at least two depth domain interpretation fault data, two depth domain interpretation level data, two time domain interpretation level data, and one time domain seismic data body.

[0158] The three-dimensional structural framework model building module 20 is used to build a three-dimensional structural framework model using the seismic result data.

[0159] In the subsequent steps, in order to analyze multiple types of data such as well information, seismic data, and logging interpretation data together, a three-dimensional structural framework grid model needs to be established to accommodate multiple types of data. In practical applications, the three-dimensional structural framework model is constructed by the two depth domain interpretation levels and two depth domain interpretation faults, for example, the top and bottom interfaces and faults of a certain section of an oil layer in a certain place are presented in three-dimensional form.

[0160] The data volume conversion module 30 is used to convert the time domain seismic data volume into a depth domain seismic data volume and update the well seismic database.

[0161] Since the propagation of seismic waves in the strata is inevitably affected by the properties of the strata (lithology, physical properties, fluid properties, etc.), the difference in the properties of the underground reservoir will lead to differences in the wave impedance of the reservoir, which in turn causes seismic response. Seismic attributes such as frequency, amplitude, phase, and wave impedance may reflect geological characteristics such as lithology, lithofacies, rock continuity, and porosity. Based on this principle, it is necessary to analyze the relationship between seismic attributes and lithofacies, porosity, and permeability, and combine the geological research results data of the study area to select multiple seismic attributes that can reflect the properties of the reservoir. However, in actual applications, the time domain seismic data body is time domain data, and the well information and geological research results data are all depth domains, so it is necessary to perform time-depth conversion on the time domain seismic data body and convert it into a depth domain seismic data body, and then extract the depth domain seismic attributes, which is convenient for analyzing the relationship between the depth domain seismic attributes and lithofacies, porosity, and permeability.

[0162] The target seismic attribute extraction module 40 is used to extract target seismic attributes from the updated well-seismic database according to preset rules, wherein the target seismic attributes can reflect reservoir changes.

[0163] Specifically, it is first necessary to extract data on seismic attributes related to lithofacies, porosity and permeability. In some embodiments, the seismic attributes related to lithofacies, porosity and permeability are sweet spots, instantaneous frequency, root mean square amplitude, envelope, inversion, etc. Then it is necessary to normalize the data of the seismic attributes, unify the dimensions, and draw a spatial intersection diagram of seismic attributes, porosity, permeability and lithofacies. Then, based on the spatial intersection diagram, the seismic attributes used to distinguish sandstone phase and mudstone phase are screened out. Finally, among the screened seismic attributes, the attribute with the highest similarity value between the spatial variation trend and the sand body distribution map and oil-bearing area map of the geological research results data is selected as the target seismic attribute.

[0164] The sampling module 50 is used to sample the target seismic attribute into the three-dimensional structural framework model and assign seismic attribute values ​​to the three-dimensional structural framework model.

[0165] In some embodiments, the sweet spot, root mean square amplitude, inversion, and instantaneous frequency seismic attribute data bodies are extracted into a three-dimensional structural framework grid model, that is, the seismic attribute data bodies are sampled into the three-dimensional structural framework grid, and seismic attribute values ​​are assigned to the three-dimensional structural framework grid. The seismic attributes are assigned to the three-dimensional structural framework grid, which can well reflect the changes in lithofacies in space.

[0166] The sedimentation unit model building module 60 is used to perform multi-attribute clustering and fusion on the seismic attribute values ​​of the three-dimensional structural framework model using a preset neural network model to build a sedimentation unit model.

[0167] The inventors hope to fuse the four seismic attributes of sweet spot, root mean square amplitude, inversion, and instantaneous frequency earthquake through an unsupervised artificial neural network and output a sedimentary unit model in order to discover the potential relationship between the attributes.

[0168] The inventors designed a neural network model and selected an autoencoder architecture as a neural network, including an encoder and a decoder. An encoder consisting of a fully connected layer was designed to map input features to a low-dimensional representation.

[0169] Let the input feature be x, and the encoder maps the input feature to a low-dimensional representation z. Assuming that the first-layer weight matrix of the encoder is W1, the bias is b1, the second-layer weight matrix is ​​W2, and the bias is b2, the output of the encoder can be expressed as:

[0170] h1=Relu(W1x+b1)

[0171] z=W2h1+b2

[0172] Design the decoder part to decode the low-dimensional representation into the original features. The structure of the decoder is opposite to that of the encoder, and a fully connected layer can be used. The decoder decodes the low-dimensional representation z into the reconstructed feature x'. Assuming that the first-layer weight matrix of the decoder is W3, the bias is b3, and the second-layer weight matrix is ​​W4, the bias is b4, then the output of the decoder can be expressed as:

[0173] h2=Relu(W3z+b3)

[0174] x'=W4h2+b4

[0175] Network training: To train the network, define a loss function that measures the error between the reconstructed output and the original input:

[0176]

[0177] Where N is the number of samples, xi is the original input of the i-th sample, and xi′ is the corresponding reconstructed output.

[0178] The main goal of neural network training is to minimize the loss function, which measures the gap between the model's output and the actual target, that is, to minimize the difference between the input data and the reconstructed data.

[0179] After the training is completed, the output sedimentary unit model is compared with the well information and geological research results data. When the conformity between the output sedimentary unit model and the well information and geological research results data reaches a preset value, the adjustment is stopped to obtain a determined sedimentary unit model.

[0180] The variogram module 70 is used to calculate the variogram between the sandstone content data for each partition of the sedimentary unit model, and estimate the sandstone content data of the unsampled position of the sedimentary unit model.

[0181] The variogram between sandstone content data is calculated for each partition of the sedimentary unit model. This function is used to describe how the variability between data points changes with increasing distance. It is usually used for spatial interpolation of geological data to understand the spatial variability of data.

[0182] It is necessary to collect sandstone content data, assign each data point to a three-dimensional spatial model, and calculate the distance between each pair of data points. If the spatial direction between the data points affects the variability, you can consider calculating the direction angle. For each pair of data points, calculate the semi-variance between them. The semi-variance represents the spatial correlation or difference between the data points. Group the semi-variance according to the distance bandwidth. The distance bandwidth is a custom parameter, indicating that the semi-variance within different distance ranges should be divided into different groups. Calculate the mean or variation of the semi-variance within each distance bandwidth. These values ​​constitute the different lags of the variogram. Fit the variogram model and use the semi-variance model (which can be an exponential model, a Gaussian model, a spherical model, etc.) to describe the change of the semi-variance with distance. Model fitting can show the characteristics of spatial variability, such as the degree of variation, range, etc. Use the fitted variogram to perform spatial interpolation and estimate the values ​​of unsampled locations.

[0183] The three-dimensional sandstone probability volume model generation module 80 is used to interpolate the known sandstone content distribution data in the variogram and the logging interpretation data on the three-dimensional structural framework grid using the Kriging algorithm to predict the sandstone content of the unknown point and obtain the three-dimensional sandstone probability volume model.

[0184] The Kriging algorithm is used to interpolate the known sandstone content distribution data in the variogram and logging interpretation results data on the 3D structural framework grid to predict the sandstone content of the unknown points. For each grid cell, the probability distribution of the Kriging prediction value of the surrounding points is calculated. The established 3D sandstone probability model is visualized, and color coding is used to represent different probability areas to intuitively understand the uncertainty of the sandstone content prediction of the unknown points.

[0185] For each unknown point of sandstone distribution within each 3D model grid cell, the Kriging interpolation method is used for prediction. Kriging is a geostatistical method used to estimate the value of an unknown location, taking into account the weight and distance of known points. In this step, for each unknown point, the Kriging predicted values ​​of its surrounding known points are used to estimate the sandstone content.

[0186] The three-dimensional sandstone probability volume established by the present invention integrates multiple types of seismic attributes and lithofacies data, and the probability value of each grid is 0 to 1. The three-dimensional sandstone probability volume can be applied as a trend constraint body to lithofacies and porosity and permeability modeling to improve the accuracy of reservoir modeling.

[0187] The more specific working contents of the device for establishing the three-dimensional sandstone probabilistic volume model of the present invention can be found in the embodiments, which will not be described in detail here.

[0188] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned method for characterizing prior information of physical property parameters when executing the computer program.

[0189] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program for executing the above-mentioned method for characterizing prior information of gas physical property parameters.

[0190] In the above detailed description, various features are grouped together in a single embodiment to simplify the disclosure. This method of disclosure should not be interpreted as reflecting an intention that the embodiments of the claimed subject matter require more features than are clearly stated in each claim. On the contrary, as reflected in the appended claims, the invention is in a state of having less than all the features of the disclosed individual embodiments. Therefore, the appended claims are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment of the invention.

[0191] The above description includes examples of one or more embodiments. Of course, it is not possible to describe all possible combinations of components or methods for the purpose of describing the above embodiments, but it should be recognized by those skilled in the art that the various embodiments may be further combined and arranged. Therefore, the embodiments described herein are intended to cover all such changes, modifications and variations that fall within the scope of protection of the appended claims. In addition, with respect to the term "comprising" used in the specification or claims, the word is encompassed in a manner similar to the term "including", as explained by "including," used as a transitional word in the claims. In addition, any term "or" used in the specification of the claims is intended to mean "non-exclusive or".

Claims

1. A method for establishing a three-dimensional sandstone probabilistic volume model, characterized in that: The following steps are involved: An integrated well-seismic database, wherein the data of the well-seismic database includes at least well information of the study area, geological research results data, seismic results data, and well logging interpretation results data, wherein the seismic results data includes at least two depth domain interpretation fault data, two depth domain interpretation layer data, two time domain interpretation layer data, and one time domain seismic data volume; Using the earthquake results data to construct a three-dimensional structural framework model; Converting the time-domain seismic data volume into a depth-domain seismic data volume, and updating the well-seismic database; Extracting target seismic attributes from the updated well-seismic database according to preset rules, wherein the target seismic attributes can reflect reservoir changes; Sampling the target seismic attribute to the three-dimensional structural framework model, and assigning seismic attribute values ​​to the three-dimensional structural framework model; The preset neural network model is used to perform multi-attribute clustering and fusion on the seismic attribute values ​​of the three-dimensional structural framework model to establish a sedimentary unit model; The variogram between the sandstone content data is calculated for each partition of the sedimentary unit model, and the sandstone content data of the unsampled position of the sedimentary unit model is estimated; According to the known sandstone content distribution data in the variogram and logging interpretation data, the Kriging algorithm is used to interpolate on the three-dimensional structural framework grid to predict the sandstone content of unknown points and obtain a three-dimensional sandstone probability volume model.

2. The method of establishing as claimed in claim 1, characterized in that: Using the seismic data to construct a three-dimensional structural framework model, including: The three-dimensional structural framework model is constructed by the two depth domain interpretation levels and the two depth domain interpretation faults.

3. The establishment method according to claim 1, characterized in that: Converting the time domain seismic data volume into depth domain seismic volume data includes: The time domain seismic data volume is converted into a depth domain seismic data volume using a preset velocity model, wherein the preset velocity model is: Among them, V1 and V2 represent two time domain interpretation level data respectively, Z represents depth, Z1 and Z2 are two depth domain interpretation level data, and the v interp It is the depth domain seismic data volume converted from the time domain seismic data volume.

4. The establishment method according to claim 1, characterized in that: In the updated well-seismic database, target seismic attributes are extracted according to preset rules, including: Extract data on seismic attributes related to lithofacies and porosity and permeability; After normalizing the data of the seismic attributes and unifying the dimensions, a spatial intersection diagram of the seismic attributes, porosity, permeability, and lithofacies is drawn; Screening out seismic attributes for distinguishing sandstone facies and mudstone facies according to the spatial intersection diagram; Among the screened seismic attributes, the attributes with the highest similarity values ​​between the spatial variation trend and the sand body distribution map and oil-bearing area map of the geological research results data are selected as the target seismic attributes.

5. The establishment method according to claim 1, characterized in that: The preset neural network model is used to perform multi-attribute clustering and fusion on the seismic attribute values ​​of the 3D structural framework model to establish a sedimentary unit model, including: The seismic attribute values ​​of the three-dimensional structural framework model are input into the preset neural network model to perform multi-attribute clustering fusion; Adjust the parameters of the preset neural network model until the loss function converges to a preset level or reaches a predetermined number of training iterations; The output sedimentary unit model is compared with the well information and geological research results data. When the conformity between the output sedimentary unit model and the well information and geological research results data reaches a preset value, the adjustment is stopped to obtain a determined sedimentary unit model.

6. The establishment method according to claim 5, characterized in that: The preset neural network model includes an encoder and a decoder, wherein the encoder is represented as: h1=Relu(W1x+b1) z=W2h1+b2 Where x is the input data, z is the low-dimensional mapping of the input data, W1 is the first-layer weight matrix of the encoder, b1 is the first bias, W2 is the second-layer weight matrix of the encoder, and b2 is the second bias; The loss function of the neural network model is: Where N is the number of samples, xi is the original input of the i-th sample, and xi′ is the corresponding reconstructed output.

7. The establishment method according to claim 1, characterized in that: The step of calculating the variogram between the sandstone content data for each partition of the sedimentary unit model and estimating the sandstone content data at the unsampled position of the sedimentary unit model comprises: Collect sandstone content data for each subarea of ​​the sedimentary unit model; Calculate the semivariance between each pair of sandstone content data; Group the semivariance values ​​according to the distance bandwidth; Calculate the mean or variation of the semivariance value within each distance bandwidth to form different lag values ​​of the variogram; Fitting the model to the variogram; The fitted variogram is used for spatial interpolation to estimate the sandstone content data at unsampled locations in the sedimentary unit model.

8. The establishment method according to claim 1, characterized in that: The method also includes visualization of the obtained three-dimensional sandstone probabilistic volume model.

9. A device for establishing a three-dimensional sandstone probability volume model, characterized in that: It includes integrated well-seismic database module, 3D structural framework model building module, data volume conversion module, target seismic attribute extraction module, sampling module, sedimentary unit model building module, variogram module, and 3D sandstone probability volume model generation module, among which: The integrated well-seismic database module is used to integrate the well-seismic database, the data of the well-seismic database at least includes well information of the study area, geological research results data, seismic results data, and well logging interpretation results data, wherein the seismic results data at least includes two depth domain interpretation fault data, two depth domain interpretation layer data, two time domain interpretation layer data, and one time domain seismic data volume; The three-dimensional structural framework model construction module is used to construct a three-dimensional structural framework model using the seismic result data; The data volume conversion module is used to convert the time domain seismic data volume into a depth domain seismic data volume and update the well seismic database; The target seismic attribute extraction module is used to extract target seismic attributes according to preset rules from the updated well-seismic database, wherein the target seismic attributes can reflect reservoir changes; The sampling module is used to sample the target seismic attribute to the three-dimensional structural framework model and assign seismic attribute values ​​to the three-dimensional structural framework model; The sedimentation unit model building module is used to perform multi-attribute clustering and fusion on the seismic attribute values ​​of the three-dimensional structural framework model using a preset neural network model to establish a sedimentation unit model; The variogram module is used to calculate the variogram between the sandstone content data for each partition of the sedimentary unit model, and estimate the sandstone content data of the unsampled position of the sedimentary unit model; The three-dimensional sandstone probability volume model generation module is used to interpolate on the three-dimensional structural framework grid using the Kriging algorithm according to the known sandstone content distribution data in the variogram and the logging interpretation results data, predict the sandstone content of the unknown point, and obtain the three-dimensional sandstone probability volume model.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method of any one of claims 1 to 8 is implemented.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program for executing the method according to any one of claims 1 to 8.

Images