A method and system for quantitatively representing interwell connectivity using machine learning

Abstract

The application discloses a method and system for quantitatively representing interwell connectivity by using machine learning, and belongs to the technical field of dynamic monitoring of fire-flooded heavy oil reservoirs. The method draws a well logging curve according to well logging data of a well group of a fire-flooded block, compares and analyzes the well logging curve to determine structural similarity and trend similarity of the well logging curve between injection and production wells of the fire-flooded block, and determines a comprehensive static similarity coefficient between the injection and production wells of the fire-flooded block according to the structural similarity and the trend similarity. Main control factors affecting production of a production well are determined according to engineering data of the fire-flooded block, a main feature-multivariate injection-production correlation model is constructed in combination of the main control factors and the comprehensive static similarity coefficient, and the interwell connectivity coefficient of the injection and production wells of the fire-flooded block is determined by using the main feature-multivariate injection-production correlation model. The method can efficiently solve the dynamic monitoring and analysis of the interwell connectivity between injection and production wells in the fire-flooded development process, help optimize the dynamic monitoring and adjustment of the heavy oil reservoir fire flooding, and thus improve the recovery ratio and economic benefits.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of dynamic monitoring of heavy oil reservoirs with fire flooding, and particularly to a method and system for quantitatively characterizing interwell connectivity between injection and production wells using machine learning. BACKGROUND

[0002] Fire flooding is an effective method for improving the recovery of heavy oil and heavy oil. This technology ignites part of the oil in the reservoir to form a combustion front, which transmits heat to the deep part of the reservoir, thereby reducing the viscosity of the oil and improving its flowability. Fire flooding technology can significantly improve the ultimate recovery of the reservoir, especially in high-viscosity reservoirs that cannot be effectively exploited by traditional water injection or steam flooding. During the fire flooding process, the connectivity between the injection well and the production well is one of the key factors determining the recovery. Good interwell connectivity can ensure uniform advancement of the combustion front, improve thermal efficiency, and reduce ineffective areas.

[0003] However, the geological complexity and heterogeneity in actual reservoirs make it very difficult to quantitatively characterize interwell connectivity. Although traditional geological modeling and numerical simulation methods can provide certain predictive ability, they often require a large amount of geological data and computing resources, and the accuracy of the results depends on the quality of the input data and the assumptions of the model. SUMMARY

[0004] In view of the fact that the prior art cannot quickly and accurately quantify the interwell connectivity of heavy oil reservoirs with fire flooding, the present application aims to provide a method and system for quantitatively characterizing interwell connectivity between injection and production wells using machine learning. By calculating the interwell connectivity between injection and production wells of fire flooding based on static geological data and dynamic engineering data, the dynamic monitoring and analysis of the interwell connectivity between injection and production wells during the fire flooding development process can be efficiently solved, which helps to optimize the dynamic monitoring and adjustment of heavy oil reservoirs with fire flooding, thereby improving the recovery and economic benefits.

[0005] The present application is achieved by the following technical solutions:

[0006] In a first aspect, the present application provides a method for quantitatively characterizing interwell connectivity between injection and production wells using machine learning, comprising the following processes:

[0007] Step 1. Draw a well log curve according to the well log data of the well group in the fire flooding development block, and determine the structural similarity and trend similarity of the well log curve between the injection and production wells by comparing and analyzing the well log curve. Determine the comprehensive static similarity coefficient between the injection and production wells of fire flooding according to the structural similarity and trend similarity.

[0008] Step 2. Determine the main controlling factors affecting the production of the production well according to the engineering data of the fire flooding development block, and combine the main controlling factors and the comprehensive static similarity coefficient to construct a main feature-multivariate injection-production correlation model. Determine the interwell connectivity coefficient of the injection and production wells of fire flooding using the main feature-multivariate injection-production correlation model.

[0009] Preferably, the step 1 of drawing the well logging curve comprises:

[0010] The gamma logging data, acoustic logging data and resistivity logging data of the well group of the fire flooding block are obtained, and the well logging data is preprocessed, and the well logging curve is drawn according to the preprocessed well logging data.

[0011] Preferably, the method for determining the comprehensive static similarity coefficient between the injection and production wells of the fire flooding in step 1 is as follows:

[0012] The brightness, contrast and structure of the well logging curves of each well are compared to determine the structural similarity between the well logging curves; the curve fluctuation amount of each well in the set range is compared to determine the trend similarity between the well logging curves; and the comprehensive static similarity coefficient is determined according to the average value of the structural similarity and the trend similarity.

[0013] Preferably, the method for determining the main control factor in step 2 comprises:

[0014] The importance scores of the feature variables in the engineering data are determined by using the machine learning method, the importance scores are sorted, and the top several feature variables in the sequence are taken as the main control factors.

[0015] Preferably, in step 2, a plurality of machine learning methods are used to determine the importance scores of the feature variables in the engineering data and the accuracy scores of the importance scores of the feature variables, the optimal machine learning method is determined according to the highest accuracy score, the importance scores of the feature variables output by the optimal machine learning are sorted, and the top three feature variables in the importance scores of the feature variables are taken as the main control features.

[0016] Preferably, the plurality of machine learning methods include decision tree algorithm, random forest algorithm and AdaBoost algorithm.

[0017] Preferably, the engineering data includes oil production, water cut, gas injection amount, wellhead temperature, production days, oil pressure, well spacing and open / close well condition.

[0018] The oil production is taken as the target variable, and the water cut, monthly gas injection amount, wellhead temperature, production days, oil pressure, well spacing and open / close well condition are taken as the feature variables, and the importance scores of the feature variables are calculated.

[0019] Preferably, the method for constructing the main feature-multiple injection and production correlation model is as follows:

[0020] The comprehensive static similarity coefficient and the main feature of each target production well are taken as independent variables, the monthly oil production of the target production well is taken as dependent variable, and the main feature-multiple injection and production correlation model suitable for fire flooding is constructed by using multiple linear regression method.

[0021] Preferably, the main feature-multiple injection-production correlation model is expressed as follows:

[0022]

[0023] wherein, is the injection rate of the i th injection well at the time step n; i is the injection rate of the i th injection well at the time step n; n is the injection rate of the i th injection well at the time step n; is a constant term, an injection-production imbalance coefficient; is the comprehensive static similarity coefficient between the j th production well and the i th injection well obtained in step four; is the injection rate of the i th injection well at the time step n; j is the injection rate of the i th injection well at the time step n; i is the dynamic correlation weight between the j th production well and the i th injection well; is the injection rate of the i th injection well at the time step n; j is the injection rate of the i th injection well at the time step n.

[0024] In a second aspect, the application provides a system for quantitatively representing the connectivity between injection and production wells by using machine learning, comprising:

[0025] a static coefficient module, configured to draw a well logging curve according to well logging data of a well group of a fire-flooding development block, to determine structural similarity and trend similarity of the well logging curve between injection and production wells of the fire-flooding development block by comparative analysis of the well logging curve, and to determine a comprehensive static similarity coefficient between the injection and production wells of the fire-flooding development block according to the structural similarity and the trend similarity;

[0026] an interwell connectivity representation module, configured to determine a main control factor affecting the production of a production well according to engineering data of the fire-flooding development block, to construct a main feature-multiple injection-production correlation model by combining the main control factor and the comprehensive static similarity coefficient, and to determine an interwell connectivity coefficient of the injection and production wells of the fire-flooding development block by using the main feature-multiple injection-production correlation model.

[0027] Compared with the prior art, the application has the following beneficial technical effects:

[0028] The method for quantitatively representing interwell connectivity by using machine learning provided by the application selects main features associated with the fire flooding production by using the machine learning method, including static geological feature parameters and dynamic engineering feature parameters, constructs a main feature-multiple injection-production correlation model suitable for fire flooding, can analyze the influence of geological features and engineering features on the interwell connectivity of fire flooding, can analyze the main features associated with the production between injection wells and production wells in the actual fire flooding production process, quantitatively represents the interwell connectivity of fire flooding by calculating the model parameters, avoids the influence of redundant features on the accuracy of the machine learning process through feature selection, reduces the data analysis workload, adapts to the actual production of specific well groups in the specific reservoir fire flooding process, and efficiently and accurately quantifies the interwell connectivity of fire flooding. The application solves the problem that the interwell connectivity between injection wells and production wells in the fire flooding process is difficult to be quickly and accurately quantitatively represented, can add various data in the actual production process, and real-time monitors the interwell connectivity between injection wells and production wells in fire flooding, thereby providing a decision basis for the dynamic monitoring, dynamic adjustment and safe production of fire flooding.

[0029] The application further provides a system for quantitatively representing interwell connectivity in a fire flooding reservoir based on machine learning, an electronic device and a computer storage medium, which have all the advantages of the method for quantitatively representing interwell connectivity in a fire flooding reservoir based on machine learning. BRIEF DESCRIPTION OF DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings needed in the embodiments will be briefly introduced as follows. It should be understood that the following drawings only show some embodiments of the application, and therefore should not be regarded as a limitation on the scope, and other related drawings can also be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0031] Figure 1 The flowchart of the method for quantitatively representing interwell connectivity by using machine learning of the application;

[0032] Figure 2 The feature importance calculation result graph calculated by the three methods of the application;

[0033] Figure 3 The result accuracy graph calculated by the three methods of the application;

[0034] Figure 4 The interwell connectivity schematic diagram of the application. DETAILED DESCRIPTION

[0035] To make the purposes, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some but not all of the embodiments of the present application. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations.

[0036] Therefore, the detailed description of the embodiments of the present application provided below in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0037] In order to efficiently and explicitly determine the communication condition between injection and production wells in a fire flooding reservoir, and to help dynamic monitoring and adjustment in the process of fire flooding development, the present application provides a method for quantitatively characterizing the communication between injection and production wells in a fire flooding reservoir based on machine learning. The method uses machine learning data analysis methods, comprehensively considers geological factors and engineering development factors, combines computer science and reservoir development, determines the main engineering factors related to production in the communication between injection and production wells in a fire flooding reservoir, establishes a feature correlation model suitable for the communication between injection and production wells in a fire flooding reservoir, calculates and quantifies the communication condition between injection and production wells in a fire flooding reservoir. The method is described in detail below.

[0038] A method for quantitatively characterizing the communication between injection and production wells using machine learning, comprising the following steps:

[0039] Step 1, obtaining well group logging data of a fire flooding development block, drawing logging curves according to the well group logging data, and determining the structural similarity and trend similarity of the logging curves between the injection and production wells in the fire flooding, to obtain the comprehensive static similarity coefficient between the injection and production wells in the fire flooding, specifically as follows:

[0040] S1.1, obtaining logging data, including gamma logging data, acoustic logging data, and resistivity logging data corresponding to the formation depth of each well target oil layer.

[0041] S1.2, preprocessing the logging data, including logging data outlier processing and data denoising.

[0042] S1.3, drawing gamma logging curves, acoustic logging curves, and resistivity logging curves according to the preprocessed logging data.

[0043] S1.4, comparing and analyzing the gamma logging curves, acoustic logging curves, and resistivity logging curves to determine the structural similarity and trend similarity of the logging curves between the injection and production wells in the fire flooding.

[0044] 1) Structural Similarity (SSIM) of well logging curve images is obtained by comparing the brightness, contrast and structure of well logging curve images. The larger the value is, the more similar the structures of the well logging curves of the two wells are.

[0045]

[0046] wherein, and are local windows of the two image curves; and are the average values of the two windows; and are the standard deviations of the two windows; is the covariance of the two windows; and are constant terms added for stable calculation.

[0047] 2) Trend Similarity (TSIM) of well logging curve images is obtained by comparing the curve fluctuation within a distance between the well logging curve images. The larger the value is, the more similar the trends of the well logging curves of the two wells are.

[0048]

[0049] wherein, MSE is the mean square error, which is the average value of the square of the difference between the predicted value and the true value, and is calculated by using formula (3):

[0050]

[0051] wherein, is the value of a signal of the curve in the image; is the value of a signal of the curve in another image; is the sample quantity.

[0052] S1.5, determine the comprehensive static similarity coefficient according to the structural similarity and the trend similarity.

[0053] Comprehensive static similarity coefficient is the average value of the sum of the structural similarity coefficient and the trend similarity coefficient. The larger the value is, the closer the geological conditions of the well logging curves of the two wells are.

[0054]

[0055] Step 2, obtain the engineering data of the fire flooding development block, determine the main control factors affecting the production of the production well according to the engineering data, combine the main control factors and the comprehensive static similarity coefficient, construct a main feature-multivariate injection-production correlation model, determine the engineering factor influence weight between the injection-production wells by using the main feature-multivariate injection-production correlation model, and determine the connectivity coefficient between the injection-production wells according to the engineering factor influence weight, which is specifically as follows:

[0056] S2.1 Obtain engineering data for the fire-driven development block, including monthly oil production, water cut, monthly gas injection, wellhead temperature, production days, oil pressure, well spacing, and well opening / closing status.

[0057] S2.2 Match the response of the gas injection wells with the target production wells and preprocess the engineering data of the fire-driven development block.

[0058] S2.3. Construct initial injection and mining data samples based on the preprocessed engineering data and perform normalization processing to obtain a standard sample dataset. Divide the standard sample dataset into a training set and a test set in a ratio of 8:2.

[0059] S2.4. Using monthly oil production as the target variable, water cut, monthly gas injection, wellhead temperature, production days, oil pressure, well spacing, and well opening / closing status as feature variables; employing multiple machine learning methods to determine the importance score of each feature variable and the accuracy score of the feature variable importance score, determining the optimal machine learning method based on the highest accuracy score, ranking the feature variable importance scores output by the optimal machine learning method, and selecting the top three feature variables with the highest feature variable importance scores as the main control factors.

[0060] Various machine learning methods, including decision tree algorithms, random forest algorithms, and AdaBoost algorithms, are used to perform classification algorithms, calculate the importance score of feature variables, and the accuracy score of the feature variable importance score.

[0061] 1) Calculate the importance score of each feature variable using a decision tree, as follows:

[0062]

[0063] In the formula, For all nodes of the decision tree; For the first Use features in a node The information gain during splitting is determined as follows:

[0064]

[0065] In the formula, The entropy of the current node can be calculated using formula (7); Features Each value The corresponding subset; The number of samples in the subset; This represents the total number of node samples.

[0066]

[0067] wherein, is the number of classes; is the probability that the class is

[0068] 2) Calculate the result accuracy score using decision tree, method as follows:

[0069]

[0070] wherein, is the number of true positives, the number of samples that are predicted or classified as positive class and actually are positive class; is the number of true negatives, the number of samples that are predicted or classified as negative class and actually are negative class; is the number of false positives, the number of samples that are predicted or classified as positive class but actually are negative class; is the number of false negatives, the number of samples that are predicted or classified as negative class but actually are positive class.

[0071] 3) Calculate the importance score of each feature variable using random forest, method as follows:

[0072]

[0073] wherein, is the number of trees in the random forest; is the information gain when feature is used for splitting in the th tree.

[0074] The result accuracy calculation method of random forest is consistent with the result accuracy score calculation method of decision tree, see step 2) for details.

[0075] 4) Calculate the importance score of each feature variable using AdaBoost algorithm, wherein the algorithm uses decision tree as weak classifier, method as follows:

[0076]

[0077] wherein, is the indicator function, taking 1 if feature is used in the classifier , otherwise taking 0; is the total number of weak classifiers.

[0078] The result accuracy calculation method of AdaBoost algorithm is consistent with the result accuracy calculation method of decision tree, see step 2) for details.

[0079] ​S2.5. Using the comprehensive static similarity coefficient and the main characteristics of each target production well as independent variables, and the monthly oil production of the target production well as the dependent variable, a main characteristic-multivariate injection-production correlation model suitable for fire-driven flooding is constructed using multiple linear regression. The expression is as follows:

[0080]

[0081] in, For the first i Injection well at time step n The amount injected; For constant terms, the injection-production imbalance coefficient; The comprehensive static similarity coefficient between the j-th production well and the i-th injection well obtained in step four; For the first j production wells and the first i Dynamic correlation weights between injection wells; For the first j The production volume of a production well at time step n.

[0082] S2.6 Calculate the multiple regression weight coefficients of the main feature-multivariate injection-progression correlation model. The obtained weighting coefficient is the correlation coefficient between the fire-driven injection and production wells, which can also be understood as the inter-well connectivity coefficient of the fire-driven well group.

[0083] Step 3: Repeat steps 1 and 2 to obtain the injection-production connectivity coefficients of each production well and gas injection well in the required fire-driven oil well group, draw the injection-production connectivity diagram, and clarify the inter-well connectivity status of the fire-driven injection-production wells.

[0084] This machine learning-based method for quantitatively characterizing the connectivity between injection and production wells in fire-driven oil reservoirs utilizes geological and engineering data from the reservoir development process. Through comparative analysis of well logging curves, a comprehensive static similarity coefficient is derived. Machine learning is then used to screen the main features related to production between injection and production wells in fire-driven reservoirs. Finally, a main feature-multivariate injection-production correlation model suitable for fire-driven reservoirs is constructed using the comprehensive static similarity coefficient and the main features. The connectivity coefficient between injection and production wells is calculated, accurately quantifying the connectivity status between them. This invention solves the problem of determining the connectivity between gas injection wells and corresponding production wells during fire-driven development of heavy oil reservoirs. It can accurately and quickly identify the connectivity status between injection and production wells and can update the assessment of connectivity status in real time based on actual production, providing assistance for the regulation and control of fire-driven development in heavy oil reservoirs.

[0085] Example 1

[0086] Taking a fire-driven development well group in a heavy oil block of a certain oilfield as an example, this paper describes a method of using machine learning to quantitatively characterize the connectivity between injection and production wells.

[0087] S100, collect well logging data and engineering data of an oilfield block to be analyzed, the well group well logging data and engineering data including as shown in Table 1.

[0088] Table 1

[0089]

[0090] S200: draw a well logging curve using the well logging data of the target oil layer steam injection well and the target production well collected in S100.

[0091] S300: perform noise reduction processing on the well logging curve obtained in S200.

[0092] S400: calculate the structural similarity and the trend similarity of the well logging curve of the target oil layer steam injection well and the well logging curve image of the target production well, and obtain a comprehensive static similarity coefficient of the steam injection well and the target oil well, as shown in Table 2.

[0093] Table 2

[0094]

[0095] S500: match the response of the steam injection well and the target production well, perform data preprocessing on the engineering data obtained in S100, and the preprocessed data constitutes an initial injection-production data sample. Normalize the initial injection-production sample data to form a standard sample data set, and divide the standard sample data set into a training set and a test set according to a ratio of 8:2.

[0096] S600: rely on the training set, take the monthly oil production as the target variable, take the water cut, monthly steam injection volume, wellhead temperature, production days, oil pressure, well spacing and open / close well condition as the characteristic variables, and perform feature importance score calculation and result accuracy calculation based on decision tree, random forest and AdaBoost algorithm respectively. The feature importance score calculation result is as shown in Figure 2 , and the result accuracy calculation result is as shown in Figure 3 .

[0097] S700: compare and analyze the result accuracy of S600, select the algorithm with the highest result accuracy value as the algorithm for calculating the feature importance score result, and obtain a feature importance ranking. The top three in the feature importance ranking (ranked from high to low according to the feature importance score) are the main features affecting the monthly oil production of the target production well, and Figure 2 , Figure 3 The main feature selection is water cut, monthly steam injection and oil pressure.

[0098] S800: selecting the comprehensive static similarity coefficient obtained in S400 and the main features of each target production well in S700 as independent variables, and the monthly oil production of the target production well as a dependent variable, to construct a main feature-multiple injection-production correlation model suitable for fire flooding;

[0099] S900: calculating the multiple regression weight coefficient of the main feature-multiple injection-production correlation model, and the obtained weight coefficient is the correlation coefficient between the injection-production wells of the fire flooding, which is also the connectivity coefficient between the injection-production wells of the fire flooding, and the result is shown in Table 3.

[0100] Table 3

[0101]

[0102] S1000: repeating steps S100 to S900 to obtain the injection-production connectivity coefficients of each production well and gas injection well in the required fire flooding oil well group, drawing an injection-production connectivity diagram, and clearly showing the injection-production well interconnection condition of the fire flooding, as shown in Figure 4

[0103] Based on the above method for quantitatively representing the injection-production well interconnection by using machine learning, the application also provides a system for quantitatively representing the injection-production well interconnection by using machine learning, which comprises:

[0104] a static coefficient module, configured to draw a well logging curve according to well logging data of a fire flooding development block well group, to determine the structural similarity and the trend similarity of the well logging curve between the injection-production wells of the fire flooding by comparative analysis on the well logging curve, and to determine the comprehensive static similarity coefficient between the injection-production wells of the fire flooding according to the structural similarity and the trend similarity;

[0105] an interwell connectivity representation module, configured to determine the main control factors affecting the production well yield according to engineering data of the fire flooding development block, to construct a main feature-multiple injection-production correlation model in combination with the main control factors and the comprehensive static similarity coefficient, and to determine the interwell connectivity coefficient of the injection-production wells of the fire flooding by using the main feature-multiple injection-production correlation model.

[0106] It should be noted that in the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative, for example, the division of the modules is only a logical function division, and actual implementation can have another division manner, for example, multiple modules can be combined or integrated into another device, or some features can be ignored or not executed. The modules described as separate components can be or can not be physically separated, and the components displayed as modules can be one physical unit or multiple physical units, that is, they can be located in one place or distributed to multiple different places. According to actual needs, part or all of the modules can be selected to achieve the purpose of the embodiment scheme.​

[0107] In addition, each module in various embodiments of the present application can be integrated in one processing unit, or each module can exist physically separately, or two or more modules can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0108] The electronic device provided in the embodiments of the present application includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the method for quantitatively representing the interwell connectivity of a fire-flooded reservoir based on machine learning as described in any of the above embodiments when executing the computer program.

[0109] The electronic device provided in another embodiment of the present application can further include: an input port connected to the processor, configured to transmit the multi-modal data collected by an external collection device to the processor; a display unit connected to the processor, configured to display the processing result of the processor to the outside world; and a communication module connected to the processor, configured to realize the communication between the electronic device and the outside world. The display unit can be a display panel, a laser scanning display, etc. The communication mode adopted by the communication module includes but is not limited to mobile high-definition link technology (HML), universal serial bus (USB), high-definition multimedia interface (HDMI), wireless connection (including wireless fidelity technology (WiFi), Bluetooth communication technology, low-power Bluetooth communication technology, and IEEE 802.11s-based communication technology).

[0110] The computer readable storage medium provided in the embodiments of the present application stores a computer program, and the computer program is executed by the processor to implement the steps of the method for quantitatively representing the interwell connectivity of a fire-flooded reservoir based on machine learning as described in any of the above embodiments.

[0111] The related parts of the system for quantitatively representing the interwell connectivity of a fire-flooded reservoir based on machine learning, the electronic device and the computer readable storage medium provided in the embodiments of the present application are described in detail in the corresponding part of the method for quantitatively representing the interwell connectivity of a fire-flooded reservoir based on machine learning provided in the embodiments of the present application, and will not be described here. In addition, the parts of the above technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail, so as not to be too repetitive.

[0112] The above content only illustrates the technical idea of the present application, and cannot limit the protection scope of the present application. Any modification made according to the technical idea of the present application on the basis of the technical solutions falls within the protection scope of the claims of the present application.

Claims

1. A method for quantifying characterization of interwell connectivity using machine learning, the method comprising: The method comprises the following steps: Step 1: According to the well logging data of the fire-flooded block well group, the well logging curves are drawn, the structural similarity and trend similarity of the well logging curves between the injection and production wells of the fire flooding are determined by comparing and analyzing the well logging curves, and the comprehensive static similarity coefficient between the injection and production wells of the fire flooding is determined according to the structural similarity and the trend similarity. The determination method of the comprehensive static similarity coefficient is as follows: The structural similarity between the well logging curves is determined by comparing the brightness, contrast and structure of the well logging curves of each well; and the trend similarity between the well logging curves is determined by comparing the curve fluctuation amount of each well in a set range. The comprehensive static similarity coefficient is determined according to the average value of the structural similarity and the trend similarity. Step 2: The main control factors affecting the production of the production well are determined according to the engineering data of the fire-flooded block, and a main feature-multivariate injection-production correlation model is constructed by combining the main control factors and the comprehensive static similarity coefficient, so as to determine the interwell connectivity coefficient of the injection and production wells of the fire flooding. The determination method of the main control factors comprises: The importance scores of the feature variables in the engineering data are determined by using a machine learning method, the importance scores are sorted, and the top several feature variables in the sequence are taken as the main control factors, which are as follows: The importance scores of the feature variables in the engineering data and the accuracy scores of the importance scores of the feature variables are determined by using a plurality of machine learning methods, the optimal machine learning method is determined according to the highest accuracy score, and the importance scores of the feature variables output by the optimal machine learning are sorted, and the top three feature variables are taken as the main control features. The construction method of the main feature-multivariate injection-production correlation model is as follows: The comprehensive static similarity coefficient and the main control factors of each target production well are taken as independent variables, and the monthly oil production of the target production well is taken as a dependent variable, and a main feature-multivariate injection-production correlation model suitable for the fire flooding is constructed by using a multivariate linear regression method. The expression of the main feature-multivariate injection-production correlation model is as follows: wherein, is the constant term, the injection-production imbalance coefficient; i is the injection rate of the i-th injection well at the time step n; n is the injection rate of the i-th injection well at the time step n; is the constant term, the injection-production imbalance coefficient; is the comprehensive static similarity coefficient between the j-th production well and the i-th injection well; is the injection rate of the i-th injection well at the time step n; j is the injection rate of the i-th injection well at the time step n; i is the dynamic correlation weight between the j-th production well and the i-th injection well; is the injection rate of the i-th injection well at the time step n; j is the liquid production rate of the j-th production well at the time step n.

2. The method for quantifying interwell connectivity using machine learning according to claim 1, wherein, The well logging curves drawn in step 1 comprise: The gamma logging data, acoustic logging data and resistivity logging data of the fire-flooded block well group are obtained, the logging data are preprocessed, and the well logging curves are drawn according to the preprocessed logging data.

3. The method for quantifying interwell connectivity using machine learning according to claim 1, wherein, The plurality of machine learning methods comprise a decision tree algorithm, a random forest algorithm and an AdaBoost algorithm.

4. The method for quantifying interwell connectivity using machine learning according to claim 1, wherein, The engineering data comprise oil production, water cut, gas injection amount, wellhead temperature, production days, oil pressure, well spacing and open / close well conditions; The oil production is taken as a target variable, the water cut, monthly gas injection amount, wellhead temperature, production days, oil pressure, well spacing and open / close well conditions are taken as feature variables, and the importance scores of the feature variables are calculated.

5. A system for performing the method of quantifying interwell connectivity using machine learning of any of claims 1-4, wherein, The method comprises: A static coefficient module is configured to draw well logging curves according to well logging data of a fire-flooded block well group, compare and analyze the well logging curves to determine structural similarity and trend similarity of well logging curves between injection and production wells of the fire flooding, and determine a comprehensive static similarity coefficient between the injection and production wells of the fire flooding according to the structural similarity and the trend similarity. The interwell connection characterization module is used for determining a main control factor affecting the production of the production well according to the fire flooding block engineering data, combining the main control factor and the comprehensive static similarity coefficient, constructing a main feature-multiple injection-production correlation model, and determining the interwell connection coefficient of the fire flooding injection-production well by using the main feature-multiple injection-production correlation model.

Images