Oil reservoir automatic historical fitting data processing method and device

Abstract

The invention provides an automatic historical fitting data processing method and device for an oil reservoir. A grid structure of a target area where the target well is located, a prior oil reservoir parameter set and simulation observation data corresponding to the prior oil reservoir parameter set are obtained; according to the spatial position of the target well in the grid structure, the Euclidean distance from the center of each grid unit to the target well is calculated, and according to the Euclidean distance, a distance field corresponding to the grid structure is determined; determining a localization matrix corresponding to the grid structure according to the distance field and a preset localization function; according to the localization matrix, spatial weighted correction is carried out on a covariance matrix determined based on the prior oil reservoir parameter set and the simulation observation data corresponding to the prior oil reservoir parameter set, and a covariance matrix subjected to spatial localization correction is obtained; and according to the covariance matrix subjected to spatial localization correction, updating the priori oil reservoir parameter set to obtain a target oil reservoir parameter set. Therefore, the stability and accuracy of oil reservoir parameter fitting are improved.

Description

TECHNICAL FIELD

[0001] The present specification belongs to the technical field of oil and gas development, and particularly relates to an oil reservoir automatic history matching data processing method and device. BACKGROUND

[0002] In the prior art, the fitting of oil reservoir parameters is usually based on global calculation of a covariance matrix, which has the problems of unstable fitting and poor accuracy.

[0003] At present, no effective solution has been proposed for the above problems. SUMMARY

[0004] The present specification provides an oil reservoir automatic history matching data processing method and device, which introduces spatial localization processing to solve the problems of unstable fitting and poor accuracy caused by long-distance pseudo-correlation in the prior art.

[0005] The present specification provides an oil reservoir automatic history matching data processing method, comprising:

[0006] obtaining a grid structure of a target area where a target well is located, a set of prior oil reservoir parameters, and simulated observation data corresponding to the set of prior oil reservoir parameters;

[0007] According to the spatial position of the target well in the grid structure, the Euclidean distance from the center of each grid cell to the target well is calculated, and according to the Euclidean distance, a distance field corresponding to the grid structure is determined;

[0008] According to the distance field and a preset localization function, a localization matrix corresponding to the grid structure is determined;

[0009] According to the localization matrix, a covariance matrix determined based on the set of prior oil reservoir parameters and the simulated observation data corresponding to the set of prior oil reservoir parameters is spatially weighted and corrected to obtain a spatially localized and corrected covariance matrix;

[0010] According to the spatially localized and corrected covariance matrix, the set of prior oil reservoir parameters is updated to obtain a set of target oil reservoir parameters.

[0011] In one embodiment, the spatially weighted correction of the covariance matrix based on the set of prior oil reservoir parameters and the simulated observation data corresponding to the set of prior oil reservoir parameters according to the localization matrix to obtain a spatially localized and corrected covariance matrix comprises:

[0012] According to the set of prior oil reservoir parameters and the simulated observation data corresponding to the set of prior oil reservoir parameters, a parameter covariance matrix and a parameter-observation covariance matrix are determined;

[0013] determining spatial weighting factors corresponding to each position element of the parameter-observation covariance matrix according to spatial weight values in the localization matrix corresponding to each grid cell;

[0014] weighting each position element of the parameter-observation covariance matrix according to the spatial weighting factors to determine a localized parameter-observation covariance matrix;

[0015] determining a spatially localized covariance matrix according to the localized parameter-observation covariance matrix and the parameter covariance matrix.

[0016] In one embodiment, the determining a spatially localized covariance matrix according to the localized parameter-observation covariance matrix and the parameter covariance matrix comprises:

[0017] determining a spatially localized covariance matrix according to the following formula:

[0018]

[0019] wherein, represents an updated reservoir parameter vector of the jth sample, represents a prior reservoir parameter vector of the jth sample, represents a spatial weight matrix composed of localization functions, represents a parameter-observation covariance matrix, represents a parameter covariance matrix, represents an observation error covariance matrix, represents a random perturbation observation data generated by the jth sample, represents a simulated observation value of the jth sample.

[0020] In one embodiment, the determining a parameter covariance matrix and a parameter-observation covariance matrix according to the set of prior reservoir parameters and simulated observation data corresponding to the set of prior reservoir parameters comprises:

[0021] calculating a mean parameter vector of the reservoir parameter samples according to the plurality of reservoir parameter samples in the set of prior reservoir parameters;

[0022] calculating a mean observation vector of the observation data samples according to the plurality of observation data samples in the simulated observation data;

[0023] determining a parameter deviation vector according to differences between each reservoir parameter sample and the mean parameter vector;

[0024] determining an observation deviation vector according to differences between each observation data sample and the mean observation vector;

[0025] determine a parameter covariance matrix according to a product relationship between the parameter deviation vectors;

[0026] determine a parameter-observation covariance matrix according to a product relationship between the parameter deviation vectors and the observation deviation vectors.

[0027] In one embodiment, the determining, according to the distance field and a preset localization function, a localization matrix corresponding to the grid structure, comprises:

[0028] determining a spatial position parameter of each grid cell according to a Euclidean distance value corresponding to a position of each grid cell in the distance field;

[0029] determining a spatial weight value corresponding to each grid cell according to the spatial position parameter and a preset localization function;

[0030] determining a localization matrix corresponding to the position of the grid structure according to the spatial weight value corresponding to each grid cell.

[0031] In one embodiment, the updating, according to the spatially localized corrected covariance matrix, the prior reservoir parameter set to obtain a target reservoir parameter set, comprises:

[0032] determining a parameter update vector according to the spatially localized corrected covariance matrix, the historical observation data of the target well, and the simulated observation data;

[0033] updating the prior reservoir parameter set according to the parameter update vector to obtain a target reservoir parameter set.

[0034] In one embodiment, the updating, according to the parameter update vector, the prior reservoir parameter set to obtain a target reservoir parameter set, comprises:

[0035] determining a corresponding reservoir parameter adjustment value according to the parameter update vector;

[0036] updating the prior reservoir parameter set according to the reservoir parameter adjustment value to obtain an updated reservoir parameter set;

[0037] determining a new parameter update vector according to the updated reservoir parameter set and the corresponding simulated observation data;

[0038] continuing to update the updated reservoir parameter set according to the new parameter update vector until a preset condition is met to obtain a target reservoir parameter set.

[0039] In one embodiment, the method further comprises:

[0040] According to the target reservoir parameter set, a reservoir property distribution parameter corresponding to the target region is determined.

[0041] According to the reservoir property distribution parameter, a three-dimensional reservoir model of the target region is established.

[0042] According to the three-dimensional reservoir model, a reservoir development scheme of the target region is determined; wherein the reservoir development scheme includes injection-production well arrangement structure and injection-production parameter configuration strategy.

[0043] The present specification provides an oil reservoir automatic history matching data processing device, comprising:

[0044] A data acquisition module is configured to acquire a grid structure of a target region where a target well is located, a set of prior reservoir parameters, and simulation observation data corresponding to the set of prior reservoir parameters.

[0045] A distance field determination module is configured to calculate the Euclidean distance from the center of each grid cell to the target well according to the spatial position of the target well in the grid structure, and determine a distance field corresponding to the grid structure according to the Euclidean distance.

[0046] A matrix determination module is configured to determine a localization matrix corresponding to the grid structure according to the distance field and a preset localization function.

[0047] A correction processing module is configured to perform spatially weighted correction on a covariance matrix determined based on the set of prior reservoir parameters and the simulation observation data corresponding to the set of prior reservoir parameters according to the localization matrix, to obtain a spatially localized corrected covariance matrix.

[0048] A parameter determination module is configured to update the set of prior reservoir parameters according to the spatially localized corrected covariance matrix, to obtain a target reservoir parameter set.

[0049] The present specification also provides an electronic device comprising a processor and a memory for storing processor-executable instructions, wherein the processor implements an oil reservoir automatic history matching data processing method when executing the instructions.

[0050] The present specification also provides a computer-readable storage medium having computer instructions stored thereon, wherein the instructions implement an oil reservoir automatic history matching data processing method when executed.

[0051] Based on the oil reservoir automatic history matching data processing method provided in the specification, the grid structure of the target area where the target well is located, the prior oil reservoir parameter set and the simulated observation data corresponding to the prior oil reservoir parameter set are obtained; according to the spatial position of the target well in the grid structure, the Euclidean distance from the center of each grid unit to the target well is calculated, and the distance field corresponding to the grid structure is determined according to the Euclidean distance; according to the distance field and the preset localization function, the localization matrix corresponding to the grid structure is determined; according to the localization matrix, the spatially weighted correction of the covariance matrix determined based on the prior oil reservoir parameter set and the simulated observation data corresponding to the prior oil reservoir parameter set is performed, and the spatially localized corrected covariance matrix is obtained; according to the spatially localized corrected covariance matrix, the prior oil reservoir parameter set is updated, and the target oil reservoir parameter set is obtained. In this way, by obtaining the spatial position of the target well in the grid structure and constructing the corresponding distance field and localization matrix, the spatially weighted correction of the covariance matrix is realized, which can effectively weaken the interference caused by the false correlation between the long-distance grids. The corrected covariance matrix is used to update the prior oil reservoir parameter set, thereby improving the spatial rationality and physical consistency of parameter estimation, reducing the divergence risk in the fitting process, and helping to improve the stability and accuracy of oil reservoir parameter fitting. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the specification, the drawings needed in the embodiments will be briefly introduced as follows. The drawings described below are only some embodiments described in the specification, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0053] Figure 1 is a flowchart of an oil reservoir automatic history matching data processing method provided by an embodiment of the specification;

[0054] Figure 2 is a schematic diagram of the structure of an electronic device provided by an embodiment of the specification;

[0055] Figure 3 is a schematic diagram of the structure of an oil reservoir automatic history matching data processing device provided by an embodiment of the specification;

[0056] Figure 4 is a schematic diagram of an automatic history matching process based on the covariance positioning ES-MDA algorithm provided by an embodiment of the specification;

[0057] Figure 5 is a schematic diagram of the construction process of the covariance localization provided by an embodiment of the specification;

[0058] Figure 6 is a PUNQ model top depth schematic diagram provided by one embodiment of the present specification;

[0059] Figure 7 is a traditional ES-MDA algorithm and improved ES-MDA algorithm based on covariance positioning posterior average permeability distribution comparison schematic diagram provided by one embodiment of the present specification. DETAILED DESCRIPTION

[0060] In order to enable the person skilled in the art to better understand the technical solutions in the present specification, the technical solutions in the present specification will be described clearly and completely in the following with reference to the drawings in the embodiments of the present specification. Obviously, the described embodiments are only part of the embodiments of the present specification, not all. Based on the embodiments in the present specification, all other embodiments obtained by the person skilled in the art without creative labor should belong to the scope of protection of the present specification.

[0061] Reference Figure 1 It is shown that the present specification provides an oil reservoir automatic history matching data processing method, wherein the method is specifically applied to the server side. In specific implementation, the method can include the following contents:

[0062] S101: acquiring a grid structure of a target area where a target well is located, a set of prior oil reservoir parameters and simulation observation data corresponding to the set of prior oil reservoir parameters;

[0063] S102: calculating the Euclidean distance from the center of each grid unit to the target well according to the spatial position of the target well in the grid structure, and determining the distance field corresponding to the grid structure according to the Euclidean distance;

[0064] S103: determining the localization matrix corresponding to the grid structure according to the distance field and a preset localization function;

[0065] S104: spatially weighting and correcting the covariance matrix determined based on the set of prior oil reservoir parameters and the simulation observation data corresponding to the set of prior oil reservoir parameters according to the localization matrix, to obtain a spatially localized and corrected covariance matrix;

[0066] S105: updating the set of prior oil reservoir parameters according to the spatially localized and corrected covariance matrix, to obtain a set of target oil reservoir parameters.

[0067] The grid structure can refer to a three-dimensional discrete grid in a reservoir numerical model for describing a target area where a target well is located, including information such as spatial division, size and relative position of grid cells. The structure is used to clearly define the spatial relationship between the target well and each grid cell, and serves as the basis for constructing the distance field and the localization matrix.

[0068] The set of prior reservoir parameters can refer to a set of reservoir property parameters initially set according to geological interpretation or historical simulation experience before introducing actual observation data for fitting, including permeability, porosity, relative permeability curve, initial pressure field, saturation distribution, etc., for reflecting the initial cognition of the reservoir state.

[0069] The simulated observation data can refer to simulation results obtained by a reservoir numerical simulation tool based on the set of prior reservoir parameters, usually including simulated values of injection and production well pressure, oil production, water production, etc. over time, for difference analysis with historical observation data of the target well to guide parameter updating.

[0070] The spatial position of the target well in the grid structure can refer to the specific position of the well trajectory or wellbore coordinates of the target well in the three-dimensional grid model, which is usually determined by wellhead coordinates and inclination data, and is used to calculate the geometric distance between the well and the grid cells.

[0071] It should be noted that the spatial position of the target well in the grid structure usually corresponds to a grid cell, which serves as the position reference of the well point in space for calculating the Euclidean distance between it and other grid cells, and for constructing the covariance positioning weight function. In the covariance weighting process based on positioning, the main control grid cell is used to limit the range of the covariance matrix, thereby suppressing the pseudo correlation between grid cells far from the target well point in the parameter updating process, and improving the local consistency and physical reasonableness of the model parameters in the history matching process. In addition, when the target well passes through multiple grid cells in the three-dimensional grid structure, all grid cells passed through by the target well can be considered as its influence range according to the interaction relationship between the well trajectory and the grid. In the covariance positioning process, these well-controlled grid cells can be used as the center point together for calculating the minimum Euclidean distance between each grid cell and the well-controlled area.

[0072] The Euclidean distance can refer to the straight-line distance between the target well and the center of each grid cell in three-dimensional space, which is used to quantify the geometric proximity between the target well and the grid, and is an important basis for constructing the localization weight.

[0073] The distance field can refer to a spatial distribution field representing the Euclidean distance between the target well and all grid cells in the model, which is a spatial input quantity for constructing the localization function and reflects the decay degree of spatial correlation.

[0074] The preset localization function can be a function for converting the distance field into a spatial weight, and is usually a Gaussian function or a Gaspari-Cohn function with a spatial decay feature. The function is used to assign a smaller weight to a grid far from the target well, so as to suppress the long-range pseudo correlation.

[0075] The localization matrix can be a two-dimensional matrix generated by the localization function acting on the distance field, and each element of the matrix represents a spatial weight coefficient between the target well and the corresponding grid cell, which is used to weight and correct the false correlation caused by the spatial-independent or long-distance position in the parameter covariance matrix, so as to improve the fitting stability.

[0076] In some embodiments, the Euclidean distance from the center of each grid cell to the target well is calculated according to the spatial position of the target well in the grid structure, and the distance field corresponding to the grid structure is determined according to the Euclidean distance. Specifically, the method can include:

[0077] The grid structure information of the area where the target well is located is obtained, and the grid structure includes a plurality of three-dimensional grid cells, each grid cell having a center point spatial coordinate. The spatial position of the target well in the grid structure is determined based on the well trajectory of the target well, which can be reconstructed in three dimensions by wellhead coordinates, well depth and inclination data. For each grid cell in the grid structure, the spatial coordinates of the center point are extracted respectively, and the three-dimensional geometric distance between the coordinates and the corresponding spatial position of the target well is taken as the Euclidean distance to form a Euclidean distance set containing the distances corresponding to all grid cells. Further, the distance field is constructed based on the Euclidean distance set, and the distance field is a spatial distribution mapping of the Euclidean distance between each grid cell and the target well in the entire grid structure.

[0078] In some embodiments, the localization matrix corresponding to the grid structure is determined according to the distance field and the preset localization function. Specifically, the method can include:

[0079] Based on the spatial position of the target well in the grid structure, the distance value between the center of each grid cell in the grid structure and the target well is obtained to form a distance field. The distance field is used to represent the spatial relative relationship between each position in the grid structure and the target well.

[0080] Then, the preset localization function reflects the decay law of the spatial correlation with the distance. The localization function can be a decreasing function with a limited support range, which is used to assign a higher spatial weight to a grid cell with a shorter distance and a lower or even zero weight to a grid cell with a longer distance, so as to reflect the high correlation feature in the local space.

[0081] Then, based on the distance field and the localization function, a corresponding spatial weight coefficient is calculated for each grid cell in the grid structure, and a localization matrix is constructed according to the spatial distribution relationship of the grid structure. Each element in the localization matrix corresponds to a grid cell in the grid structure, and is used to represent the strength of the spatial correlation between the grid cell and the target well.

[0082] In addition, the generated localization matrix can also be smoothed to avoid local weight mutation problems caused by geological anomalies or boundary effects, and to improve the continuity and stability of the local weight distribution.

[0083] Finally, the localization matrix can be used in the spatial weighting process of the covariance matrix to provide a weight basis with physical space constraints for the parameter updating process, thereby effectively suppressing false correlations in unrelated regions at a distance, improving the spatial consistency of the parameter adjustment result and the physical credibility of the numerical simulation.

[0084] In some embodiments, the distance field is determined according to a preset localization function generate a global uniform localization matrix :

[0085]

[0086]

[0087] wherein: is a localization function with a radius as a parameter; is the number of grids; is a diagonal localization matrix of the model space.

[0088] In some embodiments, the spatially localized modified covariance matrix is obtained by spatially weighting and correcting the covariance matrix determined based on the set of prior reservoir parameters and the simulated observation data corresponding to the set of prior reservoir parameters according to the localization matrix. The method can further include the following contents when implemented:

[0089] S1: determining a parameter covariance matrix and a parameter-observation covariance matrix according to the set of prior reservoir parameters and the simulated observation data corresponding to the set of prior reservoir parameters;

[0090] S2: determining a spatial weighting factor corresponding to each position element of the parameter-observation covariance matrix according to the spatial weight value corresponding to each grid cell in the localization matrix;

[0091] S3: weighting each position element of the parameter-observation covariance matrix item by item according to the spatial weighting factor to determine a localized modified parameter-observation covariance matrix.

[0092] S4: determining a spatially localized modified covariance matrix according to the localized modified parameter-observation covariance matrix and the parameter covariance matrix.

[0093] Specifically, first, a set of prior reservoir parameters in a target region and simulation observation data corresponding to the set of prior reservoir parameters are obtained. The simulation observation data can include simulated responses of oil pressure, production, water cut, etc. of a target well over time.

[0094] Based on the joint distribution characteristics between the set of prior reservoir parameters and the simulation observation data, a parameter covariance matrix for representing the internal relationship between parameter changes and a parameter-observation covariance matrix for representing the joint change trend between parameters and observation data are constructed. The parameter covariance matrix is used to quantify the joint uncertainty structure between different reservoir parameters, and the parameter-observation covariance matrix is used to quantify the response degree of parameter changes to observation changes.

[0095] Next, a localization matrix constructed based on the target well position and the grid structure in the previous step is called. Each element in the localization matrix represents the spatial weight value between the corresponding grid cell and the target well. Based on the spatial weight value, a corresponding spatial weighting factor is determined for each position element in the parameter-observation covariance matrix to reflect the spatial influence degree of the parameter corresponding to the element in the vicinity of the target well.

[0096] Subsequently, using the spatial weighting factor, a term-by-term weighting operation is performed on all position elements in the parameter-observation covariance matrix, i.e., each element is multiplied by the corresponding weighting factor to obtain a localized modified parameter-observation covariance matrix. This processing can significantly suppress the interference of distant low correlation regions on the parameter adjustment results of the target well.

[0097] Finally, the above localized modified parameter-observation covariance matrix and the parameter covariance matrix are combined to construct a spatially localized covariance matrix for subsequent parameter updating. This covariance matrix retains high-confidence correlation information within the local region and effectively filters non-physical long-range correlations, which can provide a more stable and physically reasonable covariance structure basis for parameter adjustment at the target well position.

[0098] In some embodiments, the method of determining a spatially localized modified covariance matrix according to the localized modified parameter-observation covariance matrix and the parameter covariance matrix can further include the following content when implemented:

[0099] The spatially localized modified covariance matrix is determined according to the following formula:

[0100]

[0101] wherein, represents the jth sample in the updated reservoir parameter vector, represents the jth sample in the prior reservoir parameter vector, represents a spatial weight matrix composed of a localization function, represents a parameter-observation covariance matrix, represents a parameter covariance matrix, represents an observation error covariance matrix, represents the jth sample generated random disturbance observation data, represents the jth sample in the simulated observation value.

[0102] In some embodiments, the determination of the spatially localized modified covariance matrix, when implemented, can further include:

[0103] According to the set number of history quasi-contracting rounds, the corresponding inflation factor is determined, and the sum of the reciprocals of the inflation factors of each round satisfies the normalization constraint, so as to ensure the consistency of the overall error covariance weighting;

[0104] Specifically, the following formula can be used to determine:

[0105]

[0106] wherein, is the set number of data fitting, is the inflation factor.

[0107] The prior reservoir parameter set is taken as input, a preset reservoir numerical simulation module is called, a forward process is performed, and corresponding simulated observation data are generated, which are used for subsequent comparison with the observation data;

[0108] Specifically, the following formula can be used to determine:

[0109]

[0110] wherein, is the simulated value of the production dynamic data, is a reservoir numerical simulation program, is a geological parameter input into the simulation program.

[0111] In each round of history fitting, based on the true observation data, a normal disturbance term related to the inflation factor and the observation error covariance is introduced, and a disturbed observation data set is constructed to simulate the uncertainty propagation of the observation data;

[0112] Specifically, the following formula can be used to determine:

[0113]

[0114] wherein, is the measured data with added error, is the measured data, is the covariance of the measured data.

[0115] According to the simulation observation data and reservoir parameter samples in the current round, the covariance matrix between the reservoir parameters and the simulation data is calculated, which is used to quantify the dependence between the two;

[0116] Specifically, the following formula can be used to determine:

[0117]

[0118] wherein, is the mutual covariance of the updated geological parameter model and the predicted data, is the prior reservoir parameter vector of the jth sample, is the simulation observation data vector of the jth sample, is the simulation observation data vector obtained by forward simulation, is the number of parameter samples.

[0119] Based on the simulation observation data set, the covariance matrix inside is calculated, which is used to represent the statistical variability of the simulation data itself;

[0120] Specifically, the following formula can be used to determine:

[0121]

[0122] wherein, is the covariance of the predicted data.

[0123] The cross-domain covariance matrix, simulation data covariance matrix and disturbed observation data constructed as described above jointly act on the original reservoir parameter set based on the error weighting mechanism, and the updated parameter samples are obtained;

[0124] Specifically, the following formula can be used to determine:

[0125]

[0126] wherein, represents the jth sample in the updated reservoir parameter vector, represents the prior reservoir parameter vector of the jth sample, represents the spatial weight matrix composed of the localization function, represents the parameter-observation covariance matrix, represents the parameter covariance matrix, denotes an observation error covariance matrix, denotes a random perturbation observation data generated by the jth sample, denotes a simulated observation value of the jth sample.

[0127] The above steps are repeatedly performed until all the set rounds of data assimilation are completed, and finally a target reservoir parameter set that is fully matched with the historical observation data is obtained.

[0128] In some embodiments, the method, when implemented, can further include the following content:

[0129] S1: calculating a mean parameter vector of the reservoir parameter samples according to the plurality of reservoir parameter samples in the prior reservoir parameter set;

[0130] S2: calculating a mean observation vector of the observation data samples according to the plurality of observation data samples in the simulated observation data;

[0131] S3: determining a parameter deviation vector according to the difference between each reservoir parameter sample and the mean parameter vector;

[0132] S4: determining an observation deviation vector according to the difference between each observation data sample and the mean observation vector;

[0133] S5: determining a parameter covariance matrix according to the product relationship between the parameter deviation vectors;

[0134] S6: determining a parameter-observation covariance matrix according to the product relationship between the parameter deviation vector and the observation deviation vector.

[0135] Specifically, first, a prior reservoir parameter set containing a plurality of reservoir parameter samples is obtained, each reservoir parameter sample corresponding to a complete reservoir geology parameter vector including parameter values such as permeability, porosity, and saturation that represent the properties of the formation; at the same time, a simulated observation data sample corresponding to each reservoir parameter sample is obtained, the simulated observation data reflecting time series response information such as pressure, water cut, and oil production obtained by reservoir numerical simulation calculation.

[0136] Then, based on the plurality of reservoir parameter samples, the average value of each parameter dimension in each reservoir parameter sample is calculated to obtain a mean parameter vector of the reservoir parameter samples; similarly, based on the plurality of simulated observation data samples, the average value of each observation data dimension is calculated to obtain a mean observation vector of the observation data samples.

[0137] Then, a deviation value of each reservoir parameter sample relative to the mean parameter vector is calculated to form a parameter deviation vector, and a deviation value of each observation data sample relative to the mean observation vector is calculated to form an observation deviation vector.

[0138] Further, based on the cooperative variation relationship between the parameter deviation vectors, the correlation strength between the dimensions of the reservoir parameters is determined, and a parameter covariance matrix is constructed based thereon to represent the joint variation characteristics between different reservoir parameters; at the same time, based on the cooperative variation relationship between the parameter deviation vectors and the observation deviation vectors, the linear correlation degree between the reservoir parameters and the observation data is determined, and a parameter-observation covariance matrix is constructed to describe the sensitivity of the model parameters to the variation of the observation responses.

[0139] In the above process, through statistical calculation of the sample mean and the deviation, the overall correlation structure of the prior parameter set and the simulated observation data can be accurately described, thereby providing basic data support for subsequent covariance localization correction and parameter updating.

[0140] In some embodiments, the method for determining the localization matrix corresponding to the grid structure according to the distance field and a preset localization function can further include the following steps:

[0141] S1: determining the spatial position parameters of each grid cell according to the Euclidean distance values corresponding to the positions of the grid cells in the distance field;

[0142] S2: determining the spatial weight values corresponding to each grid cell according to the spatial position parameters and a preset localization function;

[0143] S3: determining the localization matrix corresponding to the positions of the grid structure according to the spatial weight values corresponding to each grid cell.

[0144] Specifically, first, the grid structure of the target region where the target well is located is obtained, and based on the spatial position of the target well in the grid structure, the Euclidean distance from the center of each grid cell to the target well is calculated to form a corresponding distance field. Each element in the distance field represents the spatial distance of the grid cell relative to the target well, which is used to represent the proximity of the target well to the position.

[0145] Then, according to the Euclidean distance values of each grid cell in the distance field, the spatial position parameters of each grid cell are extracted, which can be the spatial scale distance of the cell center relative to the target well, representing the influence degree of the target well on the constraint of the cell.

[0146] Then, the spatial position parameters are converted into corresponding spatial weight values based on a preset localization function (e.g., a Gaspari-Cohn function or a Gaussian function). The function generally assigns higher weights to grid cells close to the target well and lower or zero weights to grid cells far away from the target well according to the rule of decreasing with spatial distance, so as to weaken the long-distance pseudo-correlation.

[0147] Finally, the spatial weight values corresponding to each grid cell are organized according to their arrangement in the grid structure to construct a localization matrix corresponding to the grid structure position. The matrix, as a spatial weighting factor, can be used to perform spatial localization processing on the correlation in the covariance matrix, thereby improving the stability and geological consistency of the subsequent parameter updating process.

[0148] In some embodiments, the method for updating the set of prior reservoir parameters based on the spatially localized and corrected covariance matrix to obtain a set of target reservoir parameters can further include the following contents:

[0149] S1: determining a parameter update vector based on the spatially localized and corrected covariance matrix, the historical observation data of the target well, and the simulated observation data.

[0150] S2: updating the set of prior reservoir parameters based on the parameter update vector to obtain a set of target reservoir parameters.

[0151] Specifically, first, based on the spatially localized and corrected covariance matrix obtained in the foregoing steps, and in combination with the historical observation data of the target well and the simulated observation data corresponding to the set of prior reservoir parameters, data assimilation processing is performed to generate a parameter update vector.

[0152] The historical observation data includes monitoring data of the target well in the actual development process, such as oil production, water cut, bottom hole pressure, etc.; and the simulated observation data is simulation output data obtained by inputting the set of prior reservoir parameters into a reservoir numerical simulator.

[0153] Subsequently, by comparing the residual (i.e., error vector) between the historical observation data of the target well and the corresponding simulated observation data, and in combination with the joint statistical information in the spatially localized and corrected covariance matrix, the parameter update vector is calculated. The vector is used to indicate the direction and amplitude of adjustment of each prior reservoir parameter sample in the current assimilation step, and embodies the constraint of observation information on reservoir parameters.

[0154] Then, the parameter updating vector is applied to the original prior reservoir parameter set, and each parameter sample in the set is corrected according to a preset updating criterion, so as to form an updated reservoir parameter set, i.e., a target reservoir parameter set. The set is closer to the actual geological characteristics revealed by the historical observation data on the basis of retaining the distribution characteristics of the prior information, and has higher physical rationality and prediction reliability.

[0155] For example, in the implementation process, a set random simulation method can be used to generate a prior reservoir parameter set containing a plurality of parameter samples, each sample containing parameter values such as permeability and porosity at a plurality of spatial positions. When updating, the sample values are adjusted in proportion according to the deviation degree of the simulation response of each sample from the actual observation data and the correlation between the samples reflected by the covariance matrix, and the updating of the entire set is completed.

[0156] The process can be iteratively performed until the updated target reservoir parameter set is fully consistent with the target well historical observation data in a statistical sense.

[0157] In some embodiments, the method for iteratively updating the prior reservoir parameter set according to the parameter updating vector to obtain a target reservoir parameter set can further include the following contents in the implementation process:

[0158] S1: determining corresponding reservoir parameter adjustment values according to the parameter updating vector;

[0159] S2: updating the prior reservoir parameter set according to the reservoir parameter adjustment values to obtain an updated reservoir parameter set;

[0160] S3: determining a new parameter updating vector according to the updated reservoir parameter set and corresponding simulation observation data;

[0161] S4: continuing to update the updated reservoir parameter set according to the new parameter updating vector until a preset condition is met to obtain a target reservoir parameter set.

[0162] Specifically, first, reservoir parameter adjustment values corresponding to each reservoir parameter sample are determined based on the previously calculated parameter updating vector.

[0163] The parameter adjustment values are used to represent the magnitude of correction of each reservoir parameter sample along its perturbation direction, and the influence degree of observation information on different spatial position parameters is embodied in combination with the covariance matrix structure.

[0164] Then, the reservoir parameter adjustment values are respectively applied to the corresponding prior reservoir parameter samples to complete the first updating of the prior reservoir parameter set, so as to obtain a set of updated reservoir parameter sets.

[0165] The updating operation can be implemented in a vector weighted superposition manner, so that the parameter samples are adjusted according to the observation constraints while retaining their original spatial structure.

[0166] Then, based on the updated reservoir parameter set, a new forward reservoir numerical simulation is performed to obtain new simulation observation data, and the deviation relationship between the samples is recalculated in combination with the historical observation data of the target well, so as to determine a new parameter updating vector.

[0167] On this basis, the above updating process is repeated to continuously update the parameter set. After each round of updating, the residual index (such as mean square error, R2 coefficient, etc.) between the simulation observation data and the historical observation data can be evaluated and compared with the preset convergence condition.

[0168] When the residual index is lower than the set error threshold, or the parameter updating vector change is lower than the set fine-tuning threshold, or the number of iterations reaches the maximum limit, the iteration process is terminated, and the target reservoir parameter set that meets the historical observation information is finally obtained.

[0169] The embodiment realizes parameter updating through step-by-step iteration and multi-round convergence, so that the updated reservoir parameter set not only fits the historical observation data, but also has stronger physical consistency and prediction stability, thereby improving the credibility of the reservoir numerical model and the scientificity of the development plan.

[0170] In some embodiments, the method can further include the following when implemented:

[0171] S1: determining the reservoir physical property distribution parameters corresponding to the target region according to the target reservoir parameter set;

[0172] S2: establishing a three-dimensional reservoir model of the target region according to the reservoir physical property distribution parameters;

[0173] S3: determining a reservoir development plan of the target region according to the three-dimensional reservoir model; wherein the reservoir development plan includes injection-production well arrangement structure and injection-production parameter configuration strategy.

[0174] Specifically, the porosity, permeability, water saturation, lithology category and other parameters included in the target reservoir parameter set can be used to assign physical properties to each grid cell in the target region grid model, thereby forming a reservoir physical property distribution parameter set representing the reservoir characteristics at different spatial locations. In actual implementation, for positions with uncertain parameters, Bayesian mean or weighted expected value can be introduced for statistical interpolation to ensure the continuity of the global parameters.

[0175] The target area is divided according to the geological structure boundary, the assigned reservoir property distribution parameters are introduced in each layer, and the structure-attribute integrated three-dimensional model is constructed in combination with the formation thickness, fault boundary and lithology distribution. The model can truly reflect the geological heterogeneity and fluid migration path, and provide a basis for subsequent dynamic simulation and optimized development.

[0176] Based on the reservoir property distribution characteristics and structure boundary conditions in the three-dimensional model, the main development layer and high-yield potential area are comprehensively evaluated; on this basis, a reasonable injection-production well arrangement structure is planned, including well type, well pattern density, well location arrangement and the like. Meanwhile, corresponding injection-production parameter configuration strategies are configured in combination with the percolation capacity and reserve enrichment degree of different regions, such as injection-production ratio, water injection intensity, production rate and start-up pressure difference and the like.

[0177] Through the above steps, the whole process closed loop from the history observation data driven reservoir parameter inversion to the reservoir property modeling, three-dimensional model construction and development strategy formulation can be realized, which helps to improve the target, accuracy and economic benefit of reservoir development.

[0178] As can be seen from the above, the oil reservoir automatic history matching data processing method provided by the embodiments of the present specification obtains the grid structure of a target area where a target well is located, a set of prior reservoir parameters and simulation observation data corresponding to the set of prior reservoir parameters; according to the spatial position of the target well in the grid structure, the Euclidean distance from the center of each grid cell to the target well is calculated, and a distance field corresponding to the grid structure is determined according to the Euclidean distance; according to the distance field and a preset localization function, a localization matrix corresponding to the grid structure is determined; according to the localization matrix, a spatially weighted correction is performed on a covariance matrix determined based on the set of prior reservoir parameters and the simulation observation data corresponding to the set of prior reservoir parameters, to obtain a spatially localized corrected covariance matrix; and according to the spatially localized corrected covariance matrix, the set of prior reservoir parameters is updated to obtain a set of target reservoir parameters. In this way, by obtaining the spatial position of the target well in the grid structure and constructing the corresponding distance field and localization matrix, spatial weighted correction of the covariance matrix is realized, which can effectively weaken the interference caused by false correlation between distant grids. The corrected covariance matrix is used to update the set of prior reservoir parameters, thereby improving the spatial reasonableness and physical consistency of parameter estimation, reducing the divergence risk in the fitting process, and helping to improve the stability and accuracy of reservoir parameter fitting.

[0179] Referring to Figure 2 As shown in the figure, the embodiments of the present specification also provide a specific electronic device, wherein the electronic device includes a network communication port 201, a processor 202 and a memory 203, and the above structures are connected by internal cables so that each structure can perform specific data interaction.

[0180] The network communication port 201 can specifically obtain a grid structure of a target area where the target well is located, a set of prior reservoir parameters, and simulated observation data corresponding to the set of prior reservoir parameters.

[0181] The processor 202 can specifically be configured to calculate the Euclidean distance from the center of each grid unit to the target well according to the spatial position of the target well in the grid structure, and determine a distance field corresponding to the grid structure according to the Euclidean distance; determine a localization matrix corresponding to the grid structure according to the distance field and a preset localization function; perform spatial weighted correction on a covariance matrix determined based on the set of prior reservoir parameters and the simulated observation data corresponding to the set of prior reservoir parameters according to the localization matrix, to obtain a spatially localized and corrected covariance matrix; and update the set of prior reservoir parameters according to the spatially localized and corrected covariance matrix, to obtain a set of target reservoir parameters.

[0182] The memory 203 can specifically be configured to store corresponding instruction programs.

[0183] Based on the above method, the relevant structural performance of the electronic device can be effectively utilized, the data processing speed of the electronic device can be improved, and the oil reservoir automatic history matching data processing method can be efficiently implemented.

[0184] In this embodiment, the network communication port 201 can be a virtual port that can send or receive different data by binding with different communication protocols. For example, the network communication port can be a port responsible for web data communication, a port responsible for FTP data communication, or a port responsible for mail data communication. In addition, the network communication port can also be an entity communication interface or a communication chip. For example, it can be a wireless mobile network communication chip such as GSM, CDMA, etc.; it can also be a Wifi chip; and it can also be a Bluetooth chip.

[0185] In this embodiment, the processor 202 can be implemented in any appropriate manner. For example, the processor can take the form of, for example, a microprocessor or processor and a computer readable medium storing computer readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, an Application Specific Integrated Circuit (ASIC), a programmable logic controller, and an embedded microcontroller, etc. The present specification is not limited in this regard.

[0186] In the embodiment, the memory 203 can include a hierarchy, and can be any memory that can save binary data in a digital system; in an integrated circuit, a circuit without a physical form that has a storage function is also called a memory, such as RAM, FIFO, etc.; in a system, a storage device with a physical form is also called a memory, such as a memory stick, a TF card, etc.

[0187] The embodiment of the present specification also provides a computer readable storage medium based on the above automatic history matching data processing method for oil reservoirs, acquires a grid structure of a target region where a target well is located, a set of prior oil reservoir parameters, and simulation observation data corresponding to the set of prior oil reservoir parameters; according to a spatial position of the target well in the grid structure, calculates Euclidean distances from centers of each grid unit to the target well, and according to the Euclidean distances, determines a distance field corresponding to the grid structure; according to the distance field and a preset localization function, determines a localization matrix corresponding to the grid structure; according to the localization matrix, performs spatial weighted correction on a covariance matrix determined based on the set of prior oil reservoir parameters and the simulation observation data corresponding to the set of prior oil reservoir parameters, to obtain a spatially localized and corrected covariance matrix; and according to the spatially localized and corrected covariance matrix, updates the set of prior oil reservoir parameters, to obtain a set of target oil reservoir parameters.

[0188] In the embodiment, the storage medium includes but is not limited to a random access memory (RAM), a read-only memory (ROM), a cache, a hard disk drive (HDD), or a memory card. The memory can be used to store computer program instructions. The network communication unit can be an interface set according to a standard specified by a communication protocol, and is used for network connection communication.

[0189] In the embodiment, the program instructions stored in the computer readable storage medium specifically implement functions and effects, which can be explained by comparing with other embodiments, and will not be described here.

[0190] Reference is made to Figure 3 At a software level, the embodiment of the present specification also provides an automatic history matching data processing device for oil reservoirs, which specifically can include the following structure modules:

[0191] The data acquisition module 301 is used to acquire a grid structure of a target region where a target well is located, a set of prior oil reservoir parameters, and simulation observation data corresponding to the set of prior oil reservoir parameters;

[0192] The distance field determination module 302 is configured to calculate the Euclidean distance from the center of each grid cell to the target well according to the spatial position of the target well in the grid structure, and determine a distance field corresponding to the grid structure according to the Euclidean distance.

[0193] The matrix determination module 303 is configured to determine a localization matrix corresponding to the grid structure according to the distance field and a preset localization function.

[0194] The correction processing module 304 is configured to perform spatially localized correction on a covariance matrix determined based on the set of prior reservoir parameters and the simulated observation data corresponding to the set of prior reservoir parameters according to the localization matrix, to obtain a spatially localized corrected covariance matrix.

[0195] The parameter determination module 305 is configured to update the set of prior reservoir parameters according to the spatially localized corrected covariance matrix, to obtain a set of target reservoir parameters.

[0196] In some embodiments, the correction processing module 304 is specifically implemented as a parameter covariance determination module configured to determine a parameter covariance matrix and a parameter-observation covariance matrix according to the set of prior reservoir parameters and the simulated observation data corresponding to the set of prior reservoir parameters, determine a spatial weighting factor corresponding to each position element of the parameter-observation covariance matrix according to the spatial weight value corresponding to each grid cell in the localization matrix, perform item-by-item weighting on each position element of the parameter-observation covariance matrix according to the spatial weighting factor, and determine a localized corrected parameter-observation covariance matrix, and a covariance matrix determination module configured to determine a spatially localized corrected covariance matrix according to the localized corrected parameter-observation covariance matrix and the parameter covariance matrix.

[0197] In some embodiments, the covariance matrix determination module is specifically implemented to determine the spatially localized corrected covariance matrix according to the following formula:

[0198]

[0199] wherein, represents the updated reservoir parameter vector of the jth sample, represents the prior reservoir parameter vector of the jth sample, represents a spatial weight matrix composed of the localization function, represents the parameter-observation covariance matrix, represents the parameter covariance matrix, represents the observation error covariance matrix, represents the random perturbation observation data generated by the jth sample, an analog observation value representing the jth sample.

[0200] In some embodiments, the parameter covariance determination module, in particular implementations, calculates a mean parameter vector of the reservoir parameter samples according to the plurality of reservoir parameter samples in the set of prior reservoir parameters; calculates a mean observation vector of the observation data samples according to the plurality of observation data samples in the analog observation data; determines parameter deviation vectors according to differences between each reservoir parameter sample and the mean parameter vector; determines observation deviation vectors according to differences between each observation data sample and the mean observation vector; determines a parameter covariance matrix according to a product relationship between the parameter deviation vectors; and determines a parameter-observation covariance matrix according to a product relationship between the parameter deviation vectors and the observation deviation vectors.

[0201] In some embodiments, the matrix determination module 303, in particular implementations, determines spatial position parameters of each grid cell according to the Euclidean distance values corresponding to the positions of each grid cell in the distance field; determines spatial weight values corresponding to each grid cell according to the spatial position parameters and a preset localization function; and determines a localization matrix corresponding to the position of the grid structure according to the spatial weight values corresponding to each grid cell.

[0202] In some embodiments, the parameter determination module 305, in particular implementations, determines a parameter update vector according to the spatially localized covariance matrix, the historical observation data of the target well, and the analog observation data; and an update module that updates the set of prior reservoir parameters according to the parameter update vector to obtain a set of target reservoir parameters.

[0203] In some embodiments, the update module, in particular implementations, determines corresponding reservoir parameter adjustment values according to the parameter update vector; updates the set of prior reservoir parameters according to the reservoir parameter adjustment values to obtain an updated set of reservoir parameters; determines a new parameter update vector according to the updated set of reservoir parameters and corresponding analog observation data; and continues to update the updated set of reservoir parameters according to the new parameter update vector until a preset condition is met to obtain a set of target reservoir parameters.

[0204] In some embodiments, the apparatus, in particular implementations, further includes determining reservoir property distribution parameters corresponding to a target region according to the set of target reservoir parameters; establishing a three-dimensional reservoir model of the target region according to the reservoir property distribution parameters; and determining a reservoir development plan for the target region according to the three-dimensional reservoir model; wherein the reservoir development plan includes injection-production well arrangement structures and injection-production parameter configuration strategies.

[0205] It should be noted that the units, devices or modules and the like illustrated in the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above device is described as various modules respectively described in function. Of course, in the implementation of the present specification, the functions of each module can be implemented in the same software and / or hardware, or the modules implementing the same function can be implemented by a combination of sub-modules or sub-units, etc. The above described device embodiment is only illustrative, for example, the division of the units is only a logical function division, and in actual implementation, another division mode can be used, for example, the units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point, the coupling or direct coupling or communication connection between the units or components shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0206] As can be seen from the above, based on the oil reservoir automatic history matching data processing device provided in the embodiments of the present specification, the grid structure of the target region where the target well is located, the prior oil reservoir parameter set and the simulated observation data corresponding to the prior oil reservoir parameter set are obtained; according to the spatial position of the target well in the grid structure, the Euclidean distance from the center of each grid unit to the target well is calculated, and according to the Euclidean distance, the distance field corresponding to the grid structure is determined; according to the distance field and the preset localization function, the localization matrix corresponding to the grid structure is determined; according to the localization matrix, the spatially weighted correction is performed on the covariance matrix determined based on the prior oil reservoir parameter set and the simulated observation data corresponding to the prior oil reservoir parameter set, to obtain the spatially localized and corrected covariance matrix; and according to the spatially localized and corrected covariance matrix, the prior oil reservoir parameter set is updated to obtain the target oil reservoir parameter set.

[0207] In a specific scene example, the oil reservoir automatic history matching data processing method and device provided in the present specification can be applied, and by introducing spatial localization processing, the problem of unstable fitting and poor accuracy caused by long-distance pseudo-correlation in the prior art is solved. The specific implementation process can include the following contents.

[0208] S1: target well identification and distance field calculation.

[0209] An identification matrix representing the spatial position of the target well is constructed, wherein the corresponding value of the grid unit where the target well is located is set to 1, and the corresponding value of the remaining grid units is set to 0.

[0210] According to the identification matrix, the Euclidean distance from the center of each grid unit to the nearest target well position is calculated, and a distance field corresponding to the grid structure is generated, which is used to represent the spatial proximity of each position relative to the target well.

[0211] S2: Localized function mapping and spatial weight determination.

[0212] A preset localized function (e.g., Gaspari-Cohn kernel function) with compact support characteristics and smoothness is selected to map the distance values in the distance field to spatial weight values in the interval [0, 1], the grid cells closer to the target well have higher weights, and the weights of the far distance regions exceeding the set threshold rapidly decay to 0, and finally the corresponding localized matrix is formed.

[0213] S3: Objective function construction and Bayesian inversion framework.

[0214] Under the Bayesian framework of the negative log posterior objective function, the error residual term between the historical observation data and the simulated observation data is introduced into the optimization expression together with the constraint term of the prior reservoir parameter set.

[0215] When the first derivative of the objective function is zero, the linear relationship between the target reservoir parameter set and the prior reservoir parameter set can be derived, thereby realizing the approximate inference of the posterior parameter distribution.

[0216] S4: Spatially weighted modification of covariance matrix and parameter set update.

[0217] Based on the spatial weight values corresponding to each grid cell in the localized matrix, the parameter covariance matrix and the parameter-observation covariance matrix calculated based on the prior reservoir parameter set and the simulated observation data are element-wise weighted to suppress the false correlation between the distant regions in the physical space while preserving the local true correlation structure.

[0218] On this basis, the parameter update vector is constructed, and the prior reservoir parameter set is iteratively updated for multiple rounds to finally obtain the converged target reservoir parameter set.

[0219] S5: Multi-round data assimilation process under localized constraints.

[0220] A multi-round data assimilation mechanism (e.g., ES-MDA method) is adopted, and in each round of data assimilation, the corresponding inflation factor is introduced to update the parameter set using the historical observation data and the localized modified covariance matrix.

[0221] After each round of update, the difference between the simulated observation data and the historical observation data is re-evaluated, and a new parameter update vector is constructed based on this, and the iteration continues until the preset convergence condition is reached.

[0222] In some embodiments, referring to Figure 4 As shown, the ES-MDA automatic history matching method based on covariance localization includes the following steps:

[0223] Firstly, the assimilation cycle α and the influence radius c of the spatial localization function are set to control the iteration frequency of data assimilation and the spatial range of the localization matrix.

[0224] In each assimilation cycle, i.e., for the following steps are sequentially performed:

[0225] History matching prediction stage: based on the reservoir parameter set in the current cycle , the ECLIPSE reservoir numerical simulation program is run to calculate the simulated observation data corresponding to each simulation sample at all time steps , so as to evaluate the difference between the simulation prediction and the measured value.

[0226] Observation data adjustment stage: to the actual observation data , a white noise term is added to generate the adjusted observation data for assimilation , the expression of which is:

[0227]

[0228] wherein, is the inflation factor of the current cycle, is the square root matrix of the measurement covariance matrix R, is a random vector obtained by sampling under the standard normal distribution, used to introduce the uncertainty disturbance of the measurement error.

[0229] Parameter updating stage: according to the covariance localization correction method, the localized weighted parameter-observation covariance matrix is calculated, and the simulation data covariance and the measurement error covariance are combined to update the reservoir parameters through the following formula:

[0230]

[0231] wherein, represents the jth sample in the updated reservoir parameter vector, represents the prior reservoir parameter vector of the jth sample, represents the spatial weight matrix composed of the localization function, represents the parameter-observation covariance matrix, represents the parameter covariance matrix, represents the observation error covariance matrix, represents the random disturbance observation data generated by the jth sample, represents the simulation observation value of the jth sample.

[0232] Parameter set iteration update: the updated reservoir parameter set in the current cycle is taken as the prior parameter set for the next cycle of assimilation, i.e.:

[0233]

[0234] For the next round of assimilation calculation, a multi-round smoothing update is completed.

[0235] Assimilation convergence judgment: judge whether all historical observation data have been fully absorbed and meet the preset convergence criteria, if yes, output the final set of posterior reservoir parameters, and construct the target reservoir model.

[0236] The above embodiment fully embodies the round-by-round smoothing update strategy based on the localized covariance correction, which combines steps such as simulated residual feedback, observation disturbance processing, and local weight weighting, reduces the influence of non-physical pseudo-correlation, and improves the physical consistency and fitting accuracy of geological parameter update.

[0237] In some embodiments, referring to Figure 5 As shown, a distance field is constructed according to well location information and a localized matrix corresponding to the spatial grid is generated, including: marking the positions of monitoring wells and production wells in the target area grid to form a well location identification matrix, wherein the grid cell where the well is located is marked as 1, and the remaining area grid cells are marked as 0, as shown in (a) of Figure 5 .

[0238] Based on the well location identification matrix, the Euclidean distance between each grid cell and the nearest well location is calculated to form a complete Euclidean distance distribution map, as shown in (b) of Figure 5 . Different gray values represent different Euclidean distance sizes.

[0239] Combined with a preset localization function, the Euclidean distance value is mapped to a spatial weight value in the interval [0, 1] to obtain a theoretical curve of ρ, as shown in (c) of Figure 5 . The localization function has a compact support, and the weight value gradually decays with the increase of distance when the Euclidean distance is less than the set cutoff radius c, and the corresponding spatial weight value is 0 when the distance is greater than 2c.

[0240] According to the theoretical mapping relationship, the Euclidean distance field is converted into a spatial weight distribution map corresponding to the grid structure one by one, and the spatial ρ distribution map is constructed under different cutoff radii c, as shown in (d) of Figure 5 . The ρ value distribution under the conditions of c=1, c=2, c=3, and c=4 is shown, and the non-zero weight area significantly expands and the local influence range increases as the c value increases.

[0241] In the above manner, the combination mapping of spatial position parameters and a preset localization function is completed, and a localized matrix corresponding to the grid structure one by one is established, which can be used for subsequent covariance weighting correction and parameter update operations to effectively weaken the long-distance pseudo-correlation term.

[0242] In some embodiments, referring to Figure 6 As shown, the three-dimensional stratigraphic structure model generated by the geological modeling platform (such as Petrel or CMG Builder) is used to assign parameters to the reservoir space framework of the target area and deploy well sites. Through the obtained parameter set of the target reservoir, representative parameter values reflecting the physical properties of different stratigraphic units, such as porosity, permeability, and saturation, are extracted and mapped to each unit in the three-dimensional grid. According to the assigned reservoir model, a stratigraphic depth profile as shown in the figure is visualized, and combined with the spatial distribution of existing production wells (such as PRO-1, PRO-4, PRO-5, PRO-11, and PRO-15), the reservoir structure characteristics and physical property variation trend are analyzed.

[0243] Further, according to the physical property parameter distribution and structural morphology, factors such as interwell connectivity, displacement efficiency, and stratigraphic heterogeneity are considered to construct a reservoir development plan for the target area, including the arrangement structure of injection and production wells and the configuration strategy of injection and production parameters. For example, water injection wells are deployed in areas with high porosity and low burial depth to form an injection-production advancing front, and the density of high-permeability wells is increased in low-permeability zones to improve overall displacement efficiency. The above-mentioned plan can be used as the input basis for numerical simulation of the reservoir, supporting subsequent plan optimization and dynamic adjustment.

[0244] In some embodiments, referring to Figure 7 As shown, the comparison results of the posterior average permeability distribution on the effective grid between the traditional ES-MDA algorithm and the improved ES-MDA algorithm based on covariance localization are shown. The black dashed line in the figure represents the prior model distribution frequency, the blue histogram represents the target reference permeability distribution, the green line represents the posterior result after processing by the traditional ES-MDA method, and the red line represents the result of the improved ES-MDA method with the introduction of the covariance localization factor. As can be seen from the figure, in the high permeability area (ln(K / mD)>6), the traditional ES-MDA method has a large deviation, while the improved algorithm shows better reconstruction ability in this area, and its permeability distribution is closer to the reference target, and the frequency distribution curve has a higher coincidence degree in the high permeability area, indicating that after introducing the spatial local weighting mechanism, the misleading update caused by long-distance pseudo-correlation in the traditional algorithm is effectively suppressed, the ability to describe high permeability channels is improved, and the physical reasonableness and spatial consistency of the posterior model are enhanced.

[0245] Although the present specification provides method operational steps in the order in which the steps are performed, the order of the steps can be changed based on the underlying logic of the method. The steps of the embodiments recited in the claims can be executed in any order that is practicable and / or desirable. The order of the steps can be varied in actual implementation of the method. The steps recited in the embodiments or the figures can be executed in parallel or in series (for example, in a parallel processor or multi-threaded processing environment, or even in a distributed data processing environment). The terms "comprises", "comprising", or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the presence of additional identical or equivalent elements in the process, method, article, or apparatus that comprises the element. The terms "first", "second", and the like, define names of particular elements, and do not necessarily limit the elements to these two entities.

[0246] Those skilled in the art will also appreciate that, in addition to being embodied in a purely computer readable program code manner, the controller can be implemented in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, and embedded microcontrollers, etc. to achieve the same functions by logically programming the method steps. Therefore, such a controller can be considered as a hardware component, and the means included therein for achieving various functions can also be considered as structures within the hardware component. Alternatively, the means for achieving various functions can be considered as both a software module implementing a method and a structure within a hardware component.

[0247] From the above description of the embodiments, those skilled in the art can clearly understand that the present specification can be implemented by means of software and a necessary general hardware platform. Based on such an understanding, the technical solutions of the present specification can essentially be embodied in the form of a software product. The computer software product can be stored in a storage medium, such as a ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (which can be a personal computer, a mobile terminal, a server, or a network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0248] Although the present specification is described by means of embodiments, those skilled in the art will know that the present specification has many modifications and variations without departing from the spirit of the present specification, and it is intended that the appended claims encompass these modifications and variations without departing from the spirit of the present specification.

Claims

1. A method for processing reservoir automatic historical fitting data, characterized in that, include: Obtain the grid structure, the set of prior reservoir parameters, and the simulation observation data corresponding to the set of prior reservoir parameters for the target area where the target well is located; Based on the spatial position of the target well in the grid structure, calculate the Euclidean distance from the center of each grid cell to the target well, and determine the distance field corresponding to the grid structure based on the Euclidean distance; Based on the distance field and the preset localization function, determine the localization matrix corresponding to the grid structure; Based on the localization matrix, the covariance matrix determined based on the prior reservoir parameter set and the simulated observation data corresponding to the prior reservoir parameter set is spatially weighted and corrected to obtain the spatially localized corrected covariance matrix. The prior reservoir parameter set is updated based on the spatially localized corrected covariance matrix to obtain the target reservoir parameter set.

2. The method according to claim 1, characterized in that, The step of spatially weighting and correcting the covariance matrix determined based on the prior reservoir parameter set and the corresponding simulated observation data according to the localization matrix to obtain the spatially localized corrected covariance matrix includes: Based on the prior reservoir parameter set and the corresponding simulated observation data, determine the parameter covariance matrix and the parameter-observation covariance matrix. Based on the spatial weight values ​​corresponding to each grid cell in the localization matrix, determine the spatial weighting factors corresponding to each element of the parameter-observation covariance matrix. Based on the spatial weighting factor, each element of the parameter-observation covariance matrix is ​​weighted item by item to determine the localized corrected parameter-observation covariance matrix; Based on the localized parameter-observation covariance matrix and the parameter covariance matrix, the spatially localized covariance matrix is ​​determined.

3. The method according to claim 2, characterized in that, Determining the spatially localized covariance matrix based on the localized parameter-observation covariance matrix and the parameter covariance matrix includes: The spatially localized corrected covariance matrix is ​​determined using the following formula: in, This represents the updated reservoir parameter vector for the j-th sample. Let the vector represent the prior reservoir parameters of the j-th sample. This represents the spatial weight matrix formed by the localization functions. This represents the parameter-observation covariance matrix. Represents the parametric covariance matrix. Represents the observation error covariance matrix. This represents the random perturbation observation data generated for the j-th sample. This represents the simulated observation value of the j-th sample.

4. The method according to claim 2, characterized in that, The step of determining the parameter covariance matrix and the parameter-observation covariance matrix based on the prior reservoir parameter set and the corresponding simulated observation data includes: Calculate the mean parameter vector of the reservoir parameter samples based on multiple reservoir parameter samples in the prior reservoir parameter set; Calculate the mean observation vector of the observation data samples based on multiple observation data samples in the simulated observation data; The parameter deviation vector is determined based on the difference between each reservoir parameter sample and the mean parameter vector; The observation bias vector is determined based on the difference between each observation data sample and the mean observation vector; The parameter covariance matrix is ​​determined based on the product relationship between the parameter deviation vectors. The parameter-observation covariance matrix is ​​determined based on the product relationship between the parameter deviation vector and the observation deviation vector.

5. The method according to claim 1, characterized in that, The step of determining the localization matrix corresponding to the mesh structure based on the distance field and the preset localization function includes: Based on the Euclidean distance values ​​corresponding to the positions of each grid cell in the distance field, determine the spatial position parameters of each grid cell; Based on the spatial location parameters and the preset localization function, determine the spatial weight value corresponding to each grid cell; Based on the spatial weight values ​​corresponding to each grid cell, a localization matrix corresponding to the location of the grid structure is determined.

6. The method according to claim 1, characterized in that, The step of updating the prior reservoir parameter set based on the spatially localized corrected covariance matrix to obtain the target reservoir parameter set includes: Based on the spatially localized corrected covariance matrix, the historical observation data of the target well, and the simulated observation data, determine the parameter update vector; The prior reservoir parameter set is updated based on the parameter update vector to obtain the target reservoir parameter set.

7. The method according to claim 6, characterized in that, The step of iteratively updating the prior reservoir parameter set based on the parameter update vector to obtain the target reservoir parameter set includes: Based on the parameter update vector, determine the corresponding reservoir parameter adjustment value; Based on the reservoir parameter adjustment values, the prior reservoir parameter set is updated to obtain the updated reservoir parameter set; Based on the updated reservoir parameter set and the corresponding simulated observation data, a new parameter update vector is determined; Based on the new parameter update vector, the updated reservoir parameter set is continuously updated until the preset conditions are met, and the target reservoir parameter set is obtained.

8. The method according to claim 1, characterized in that, The method further includes: Based on the target reservoir parameter set, determine the reservoir physical property distribution parameters corresponding to the target area; Based on the reservoir physical property distribution parameters, a three-dimensional reservoir model of the target area is established; Based on the three-dimensional reservoir model, a reservoir development plan for the target area is determined; wherein, the reservoir development plan includes the layout structure of injection and production wells and the configuration strategy of injection and production parameters.

9. An automatic historical data processing device for oil reservoirs, characterized in that, include: The data acquisition module is used to acquire the grid structure of the target area where the target well is located, the set of prior reservoir parameters, and the simulated observation data corresponding to the set of prior reservoir parameters; The distance field determination module is used to calculate the Euclidean distance from the center of each grid cell to the target well based on the spatial position of the target well in the grid structure, and to determine the distance field corresponding to the grid structure based on the Euclidean distance. The matrix determination module is used to determine the localization matrix corresponding to the grid structure based on the distance field and the preset localization function. The correction processing module is used to perform spatial weighted correction on the covariance matrix determined based on the prior reservoir parameter set and the simulated observation data corresponding to the prior reservoir parameter set, according to the localization matrix, to obtain the spatially localized corrected covariance matrix. The parameter determination module is used to update the prior reservoir parameter set based on the spatially localized corrected covariance matrix to obtain the target reservoir parameter set.

10. An electronic device, characterized in that, It includes a processor and a memory for storing processor-executable instructions, wherein the processor, when executing the instructions, implements the steps of the reservoir automatic history fitting data processing method according to any one of claims 1 to 8.

Images