An evaluation method for heterogeneity differentiation of unconventional reservoir fracturing horizontal wells

Abstract

The present invention discloses a method for differential evaluation of heterogeneity of fractured horizontal wells in unconventional reservoirs, including: 1) screening sample wells through clustering and obtaining the geological parameters and engineering parameters of the sample wells; 2) standardizing the geological parameters and engineering parameters of the sample wells, and linearly combining the standardized parameters of each well into principal components through principal component analysis; 3) determining the membership degree of the principal components to the evaluation levels through the Gaussian distribution membership function, determining the principal component weights, obtaining the single-well fuzzy comprehensive score, and establishing a method for differential fuzzy comprehensive evaluation of heterogeneity of fractured horizontal wells in unconventional reservoirs; 4) substituting the horizontal section parameters of the well to be fractured into the model to obtain the fuzzy comprehensive scoring profile of the horizontal section of the well to be fractured, so as to realize the differential evaluation of heterogeneity. The present invention linearly combines the factors affecting production through principal component analysis, greatly improving the pertinence and accuracy of the model.

Description

Technical Field

[0001] The present invention relates to the field of oil and gas reservoir exploration and development, and particularly to a method for differential evaluation of heterogeneity of fractured horizontal wells in unconventional reservoirs. Background Art

[0002] With the growth of global energy demand, unconventional reservoirs have become an important replacement area for maintaining oil production. The development of horizontal well drilling technology and multi-stage hydraulic fracturing technology has broken through the limitations of low porosity and permeability of unconventional reservoirs, making it possible to develop unconventional reservoirs economically and effectively.

[0003] The study of sweet spot evaluation in unconventional reservoirs obtains the potential points for stimulation with the best production increase potential through linear or non-linear combinations of geological and engineering factors affecting production, providing strong guidance for the selection of perforation positions. However, there are many factors affecting production, some of which are correlated, and their contributions to production are different. At the same time, the main factors controlling production in different blocks are also different. Simple combinations cannot comprehensively evaluate the geological-engineering heterogeneity reflecting reservoir hydrocarbon source lithology, fracturability, and stimulation degree. In addition, after a long time of fracturing exploration and construction in developed oil and gas fields, a large amount of geological, engineering, and production data has been generated. Compared with numerical simulation results, establishing an evaluation model through the analysis and mining of such data can truly reflect the reservoir characteristics of the block, with stronger pertinence and persuasion.

[0004] Data-driven models use geological and engineering factors as input features to analyze their potential relationships with production, providing another way to deal with the problem of differential evaluation of heterogeneous reservoirs using field data. Currently, various data-driven methods such as artificial neural networks, genetic algorithms, diagnostic algorithms, and fuzzy comprehensive evaluation methods have been introduced in petroleum engineering. Given the challenging environment of downhole high temperature and high pressure where sensing devices operate, deviations in logging and mud logging data are inevitable. Therefore, the model needs to be flexible and robust enough to extract the relationship with production from a large amount of geological and engineering data, and ultimately guide the design of fracturing engineering.

[0005] In the prior art, when using a fuzzy comprehensive evaluation model, only geological parameters are used as evaluation indicators, and the weights are determined by expert experience, resulting in low accuracy and reliability of the evaluation results. To further improve the scientificity of the model, some scholars have recently introduced engineering parameters and used the grey relational method and entropy weight method to determine the weights of various factors by analyzing actual geological, engineering, and production data. However, this research method is limited by the number of parameters and cannot comprehensively reflect the geological and engineering characteristics of the reservoir. Moreover, the difference in geological characteristics between historical wells and new wells is not considered in the modeling process. In addition, related research only stays at the selection of fractured wells and does not extend the research to the evaluation of geological-engineering heterogeneity of single well fracture sections. Summary of the Invention

[0006] In view of this, the present invention takes into account the geological feature differences between the well to be fractured and historical wells, screens sample wells through geological factor clustering, and also takes into account the model noise error caused by too many factors. By linearly combining the production influencing factors through principal component analysis, the pertinence and accuracy of the model will be greatly improved. And the single-well evaluation of the well to be fractured is extended to the differential evaluation of the geological engineering heterogeneity of the horizontal section, which can better guide the design of shale oil fracturing programs.

[0007] To achieve the above object, an embodiment of the present invention provides a method for differential evaluation of heterogeneity of fractured horizontal wells in unconventional reservoirs, including the following steps:

[0008] 1) Screen historical wells with geological features similar to those of the well to be fractured as sample wells through geological factor clustering, and obtain the geological parameters and engineering parameters of the sample wells;

[0009] 2) Standardize the geological parameters and engineering parameters of the sample wells, and linearly combine the standardized parameters of each well into principal components through principal component analysis;

[0010] 3) Determine the membership degree of the principal components to the evaluation levels through the Gaussian distribution membership function, determine the principal component weights through the entropy weight method, combine the principal component membership degree matrix and the weight matrix to obtain the single-well fuzzy comprehensive score, and establish a method for differential fuzzy comprehensive evaluation of heterogeneity of fractured horizontal wells in unconventional reservoirs;

[0011] 4) Substitute the geological parameters of the horizontal section of the well to be fractured and the average engineering parameters of the sample wells into the model to obtain the fuzzy comprehensive scoring profile of the horizontal section of the well to be fractured, and realize the differential evaluation of the heterogeneity of the well to be fractured.

[0012] Further, in step (1), the sample wells are screened through the FCM clustering algorithm, and the FCM clustering algorithm is:

[0013]

[0014] where n is the number of factors in the data set x; α is the number of subsets; μ ij is the membership degree, indicating the similarity degree between the data point x i and the subset j, and the constraint condition is m is the fuzzy degree parameter; ν j is the class center value of the jth subset.

[0015] Further, the geological parameters include natural gamma, total organic carbon content, pyrolysis parameters, porosity, permeability, Young's modulus, Poisson's ratio, minimum horizontal principal stress, and horizontal principal stress difference coefficient.

[0016] Furthermore, the engineering parameters include the average section length, average cluster spacing, displacement per unit length, sand addition intensity, and fluid consumption intensity.

[0017] Furthermore, the normalization equation in step (2) is:

[0018]

[0019] In the formula, is the normalized value, x ij is the value of the production influencing factor; minx ij is the minimum value of this factor in the sample; maxx ij is the maximum value of this factor in the sample.

[0020] Furthermore, step (2) also includes the steps of:

[0021] Obtaining the eigenvalues λ 1 ≥λ 2 ≥…≥λ n ≥0 and the corresponding eigenvectors v 1 ,v 2 ,…,v n , where v j =(v 1j ,v 2j ,…,v nj ) T , v nj represents the nth component of the jth feature, and each principal component is a linear combination of the original factors:

[0022]

[0023] In the formula, z n is the nth principal component; is the normalized value of the nth factor of the nth well;

[0024] The cumulative variance contribution rate of the principal components is:

[0025]

[0026] In the formula, α k is the cumulative variance contribution rate of the first k principal components, dimensionless.

[0027] Furthermore, step (3) also includes:

[0028] Converting the fuzzy comprehensive evaluation result into an intuitive score according to the scores assigned to each level:

[0029]

[0030] In the formula, ξ is the fuzzy comprehensive score; f i represents the membership degree of the well to the i-th level; d i is the score;

[0031] Establish an evaluation matrix based on m wells and k principal components in the sample well system:

[0032]

[0033] The entropy value e of each principal component j The determination method is:

[0034]

[0035] In the formula, is the proportion of each sample point in this principal component; e j is the entropy value of the j-th principal component, dimensionless;

[0036] The entropy weight of each principal component is:

[0037]

[0038] In the formula, w j is the weight of the j-th principal component, 0 ≤ w j ≤ 1 and Thus, the principal component weight matrix can be obtained:

[0039] W = [w 1 … w j … w k 1×k (12)

[0040] In summary, the present invention has the following advantages: The improved technical solution selects historical wells with geological characteristics similar to the well to be fractured as samples through cluster analysis, uses principal component analysis to reduce the dimensions of geological and engineering factors of the sample wells to principal components, further uses the entropy weight method to determine the principal component weights according to the production of historical wells, combines the Gaussian distribution membership function and fuzzy grade division, and establishes a fuzzy comprehensive evaluation model. It makes full use of the actual geological, engineering, and production data of the block to realize the differential evaluation of the geological-engineering heterogeneity of horizontal wells in shale oil fracturing. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is the Gaussian membership function diagram of different evaluation grades in an embodiment of the present invention;

[0042] Figure 2 is the geological factor clustering result diagram in an embodiment of the present invention.

[0043] Figure 3 ​The principal component eigenvalue diagram of an embodiment of the present invention.

[0044] Figure 4 The correlation diagram between production and score of an embodiment of the present invention.

[0045] Figure 5 The geological parameter distribution diagram of an embodiment of the present invention.

[0046] Figure 6 The distribution diagram of the evaluation result of heterogeneity differentiation of an embodiment of the present invention. Detailed implementation manners

[0047] The present invention provides a method for evaluating heterogeneity differentiation of fractured horizontal wells in unconventional reservoirs, mainly including the following steps:

[0048] (1) Screening out historical wells with geological characteristics similar to those of the well to be fractured as sample wells through geological factor clustering, and obtaining the geological parameters and engineering parameters of the sample wells.

[0049] Since the geological characteristics of each well are different, selecting wells with geological conditions similar to those of the evaluation well as samples can reduce the influence caused by the geological factor gap and improve the accuracy and pertinence of the model. The FCM clustering algorithm classifies the data set into subsets based on the similarity between factors by continuously updating the membership matrix and minimizing the objective function of the class center:

[0050]

[0051] In the formula, n is the number of factors in the data set x, and x is usually represented by a matrix; α is the number of subsets; μ ij is the membership degree, indicating the similarity degree between the data point x i and the subset j, and the constraint condition is m is the fuzziness parameter (m≥1), set to 2; ν j is the class center value of the jth subset.

[0052] According to the logging, mud logging and production data, the geological, engineering and production parameters of historical wells and the average geological parameters of the horizontal section of the well to be fractured are statistically analyzed, and historical wells with geological characteristics similar to those of the well to be fractured are screened out as sample wells through geological factor clustering.

[0053] The factors affecting production can be divided into two parts: geological factors and engineering factors. To achieve the best fracturing effect, on the one hand, it is required that the hydrocarbon abundance at the perforation is high and the fluidity is good, and on the other hand, it is required to form complex fractures through fracturing to fully transform the reservoir.

[0054] Geological factors include source rock lithology and fracturability. Source rock lithology indicators reflect the storage capacity, flow capacity, and hydrocarbon content of shale, including natural gamma (GR), total organic carbon content (TOC), pyrolysis parameter (S 1 ), porosity and permeability (k). Among them, GR reflects the shale content. As GR increases, the effective pores of the reservoir decrease, resulting in a reduction in the hydrocarbon storage space, and together with reflect the size of the storage capacity. S 1 is the liquid hydrocarbon content in C 8 ~C 29 at 300 °C, which is positively correlated with TOC, and the two can intuitively reflect the hydrocarbon content in shale. The flow capacity is controlled by k, which reflects the pressure loss during the flow of shale oil.

[0055] The fracturability indicators reflect the ability of the reservoir to form complex hydraulic fractures, including Young's modulus (E), Poisson's ratio (ν), minimum horizontal principal stress (σ h ), and horizontal principal stress difference coefficient (σ dif ). E and ν are the key factors for measuring the deformation ability of shale. The larger E is, the more brittle the rock mass is, and the easier it is for hydraulic fractures to initiate and propagate, while the influence of ν is the opposite. The smaller σ h is, the longer the main fracture formed under the same construction pressure is, and the smaller σ dif is, the more conducive it is to the formation of a network of fractures.

[0056] Engineering factors reflect the degree of reservoir stimulation, including average stage length (L), average cluster spacing (S), displacement per unit length (Q m ), sand addition intensity (η s ), and fluid consumption intensity (η w ). The larger Q m and η w are, the larger the scale of the hydraulic fracture is, and the more extensive the reservoir can be communicated. On the premise of meeting the reservoir stimulation intensity, the smaller L and S are, the more fractures are formed, and the larger η w is, the higher the fracture conductivity is, and the better the reservoir stimulation effect is. According to the above analysis, the positive and negative impacts of 9 geological factors and 5 engineering factors on production are shown in Table 1.

[0057] Table 1 Classification of influencing factors

[0058]

[0059] Compared with means such as fracture morphology monitoring and fluid production profile testing, production is the most direct and easily obtainable indicator for evaluating the fracturing effect. However, due to the different geological parameters and production systems of each well, the short-term production cannot reflect the real situation. At the same time, in order to avoid the influence of the fracture length, the production is converted to per kilometer, and the oil production per kilometer in 12 months (V 12mon ) is used as the evaluation indicator.

[0060] (2) Standardize the geological parameters and engineering parameters of the sample wells, and linearly combine the standardized parameters of each well into principal components through principal component analysis to reduce the number of parameters.

[0061] In addition to the correlation with production, there is also an obvious linear relationship among various factors, that is, multicollinearity. At the same time, the noise interference and redundancy generated by the dimensionality expansion of the dataset will significantly increase the model error. Principal component analysis performs feature selection through the contribution degree of principal components, and reduces the dimensionality of the dataset and eliminates the correlation among factors without losing important information.

[0062] Due to the differences in the dimensions of the feature data, it is difficult to compare and perform weighted processing with each other. Before dimensionality reduction, the data needs to be pre-processed by standardization according to Table 1, and the standardization equation is:

[0063]

[0064] In the formula, is the standardized value, x ij is the value of the production influencing factor; minx ij is the minimum value of this factor in the sample; maxx ij is the maximum value of this factor in the sample.

[0065] The dataset containing n factors generates n mutually orthogonal and independent principal components after PCA processing. Principal component 1 always represents the direction of the maximum variance of the dataset, principal component 2 represents the second largest variance direction, and so on.

[0066] By standardizing the covariance matrix of the dataset, the eigenvalues λ 1 ≥λ 2 ≥…≥λ n ≥0 and the corresponding eigenvectors v 1 ,v 2 ,…,v n can be obtained, where v j =(v 1j ,v 2j ,…,v nj ), T v nj represents the nth component of the jth feature, and each principal component is a linear combination of the original factors:

[0067]

[0068] Wherein, z n is the nth principal component; is the standardized value of the nth factor of the nth well.

[0069] The cumulative variance contribution rate of the principal components is:

[0070]

[0071] Wherein, α k is the cumulative variance contribution rate of the first k principal components, dimensionless. The selection of the number of principal components needs to satisfy α k > 0.85, that is, when the cumulative variance contribution rate of the principal components reaches 85%, the main information of the original data set can be included.

[0072] (3) Determine the membership degree of the principal components to the evaluation level through the Gaussian distribution membership function, determine the principal component weights through the entropy weight method, combine the principal component membership degree matrix and the weight matrix to obtain the single-well fuzzy comprehensive score, and establish a differential fuzzy comprehensive evaluation method for the heterogeneity of unconventional reservoir fractured horizontal wells.

[0073] Compared with the "either-or" of Boolean logic, fuzzy logic uses membership degree to characterize the degree to which each parameter belongs to different quality levels, making each parameter have the property of "both this and that", so it is suitable for modeling non-linear functions of any complexity. The fuzzy comprehensive evaluation model includes three parts: a factor set, an evaluation set, and a weight set.

[0074] The factor set Z includes evaluation indicators and their values. The fuzzy comprehensive evaluation factor of the present invention is the principal component. The evaluation set D is used to judge the quality of the factors. To facilitate giving an intuitive score, define d 1 =Ⅰ="excellent" = 100, d 2 =Ⅱ="good" = 75, d 3 =Ⅲ="medium" = 50, d 4 =Ⅳ="poor" = 25. And the membership degree matrix R of each factor to each level can be determined through the membership function. The weight set W is the importance degree of each factor in the evaluation set determined through the entropy weight method.

[0075] Comprehensively considering the contribution of all factors to the production, combining the weight matrix W of the sample wells with the single-well principal component membership degree matrix R, the single-well heterogeneity fuzzy comprehensive evaluation matrix F can be obtained:

[0076]

[0077] Wherein, f iIndicates the membership degree of the well to the i-th level. In the present invention, g takes 4. According to the scores assigned to each level, the fuzzy comprehensive evaluation result is converted into an intuitive score:

[0078]

[0079] In the formula, ξ is the fuzzy comprehensive score.

[0080] The membership degree is jointly determined by the parameter value and the membership function. Generally, for fuzzy programming problems, linear membership functions such as triangular and trapezoidal are usually adopted, but it is no longer applicable to this multi-parameter and non-linear problem of reservoir quality evaluation. The Gaussian distribution has been favored by a large number of scholars and has achieved good application results. Therefore, the Gaussian distribution membership function is used in the present invention:

[0081]

[0082] In the formula, μ is the membership degree, μ(c, c, δ) = 1; c is the principal component value corresponding to the peak of the Gaussian distribution, dimensionless; δ is the standard deviation, dimensionless.

[0083] According to the division of the reservoir quality by the evaluation set D, the numerical ranges of the four evaluation levels corresponding to the principal components can be (0 to 0.25, 0.25 to 0.50, 0.50 to 0.75, and 0.75 to 1). According to equation (7), the Gaussian-shaped membership functions of each level can be constructed, as Figure 1 shown.

[0084] According to the membership function, the membership degree of each principal component to each level can be obtained, and further the membership degree matrix of the reservoir quality evaluation of each well can be constructed:

[0085]

[0086] Among them, r k,g is the membership degree of the k-th principal component to the g-th level.

[0087] The entropy weight method is based on the concept of information entropy. It objectively assigns weights according to the influence of the relative change degree of indicators on the overall system, can accurately reflect the internal relationship of information in the system, and further eliminates the irrationality of subjective experience judgment. Therefore, the present invention adopts this method to objectively and quantitatively determine the principal component weights according to the actual data, avoiding the errors caused by determining weights according to expert experience in traditional fuzzy logic.

[0088] Since normalization has been carried out before PCA analysis, an evaluation matrix based on m wells and k principal components in the sample well system can be directly established here:

[0089]

[0090] The entropy value e of each principal component jIt can be determined by Equation (10):

[0091]

[0092] Wherein, is the proportion of each sample point in this principal component; e j is the entropy value of the j-th principal component, dimensionless.

[0093] In the entropy weight method, the greater the degree of change of the index, the greater the weight assigned to it. The entropy weight of each principal component is:[[]]

[0094]

[0095] Wherein, w j is the weight of the j-th principal component, 0 ≤ w j ≤ 1 and Therefore, the principal component weight matrix can be established:

[0096] W = [w 1 … w j … w k 1×k (12)

[0097] (4) Substitute the geological parameters of the horizontal section of the well to be fractured and the average engineering parameters of the sample wells into the model to obtain the fuzzy comprehensive scoring profile of the horizontal section of the well to be fractured, and realize the heterogeneous differential evaluation of the well to be fractured.

[0098] Substitute the geological characteristic parameters (meter point data along the well depth) of the horizontal section of the well to be fractured obtained from well logging data and the average engineering parameters of the sample wells into the model, and the fuzzy comprehensive scoring profile of the fracturing section along the well depth can be obtained, realizing the geological engineering sweet spot differential evaluation with production as the evaluation index.

[0099] Example 1

[0100] The heterogeneous differential evaluation application was carried out for the actual well to be evaluated, Well N1 in a certain oilfield.

[0101] (1) Selection of sample wells

[0102] The sample well data comes from a certain oil reservoir, with an average reservoir depth of about 3800m. The horizontal well multi-cluster fracturing development technology is adopted in this block. Using the actual geological, engineering parameters and production of Wells Y1 - Y35 as the database (Table 2), the reservoir differential evaluation of Well N1 was carried out.

[0103] Table 2 Sample well database

[0104]

[0105] ​

[0106] To select wells with geological parameters similar to those of Well N1 as samples, the geological factors were clustered using the FCM algorithm. The number of subsets α was set to 3, and the clustering results are as follows Figure 2 shown. Among them, 11 wells belong to Class A, 7 wells belong to Class B, and 18 wells including Well N1 belong to Class C. The boundaries of each category are obvious, indicating that setting the number of clusters to 3 can achieve the classification according to the categories of geological factors.

[0107] In addition, since a large number of parameters will generate noise and affect the model accuracy, and is correlated with k and S 1 and σ h are correlated with each other, it is necessary to further carry out principal component analysis to reduce the dimension of production influencing factors and eliminate the correlation between various factors.

[0108] The results of principal component analysis are shown in Table 3. The cumulative variance contribution rate of the first 4 principal components is 87.503%, exceeding 85%. The 14 production influencing factors can be represented by 4 principal components. The eigenvalues of the principal components are as follows Figure 3 shown. According to the coefficients of each factor, the principal components can be obtained Thus, the linear combinations of the reservoir source rock lithology, fracturability, and engineering parameter indicators are combined into four comprehensive indicators.

[0109] Table 3 Results of principal component analysis

[0110] Principal component Variance contribution rate (%) Cumulative variance contribution rate (%) Principal component Variance contribution rate (%) Cumulative variance contribution rate (%) 1 30.473 30.473 8 0.924 98.624 2 24.787 55.260 9 0.632 99.256 3 19.584 74.844 10 0.414 99.670 4 12.659 87.503 11 0.178 99.848 5 4.810 92.313 12 0.091 99.939 6 3.071 95.384 13 0.057 99.996 7 2.316 97.700 14 0.004 100

[0111] According to Equation (11), the entropy weights of the first 4 principal components are calculated, and the weight matrix W = [0.3044 0.2561 0.1975 0.2421] can be obtained. The weight matrix W is combined with the membership degree matrix R of the principal components of Class C wells calculated according to Equation (7) to obtain the single-well comprehensive evaluation matrix B, and the scores of each well are calculated according to Equation (6). The fuzzy comprehensive evaluation results are shown in Table 4. Thus, a differential evaluation method for the reservoir heterogeneity of shale oil wells with geological and engineering parameters as factors and production as an index is established.

[0112] Table 4 Fuzzy comprehensive evaluation results of Class C wells

[0113]

[0114]

[0115] Figure 4 shows the fitting relationship between V 12mon and the geological-engineering score, and the correlation coefficient is 0.8877, indicating that there is a good consistency between the reservoir heterogeneity evaluation results and the actual production.

[0116] The geological data of Well N1 are as Figure 5 shown. Due to the obvious reservoir heterogeneity, it is impossible to overall consider the reservoir source rock properties and the fracturability indexes by only using one or some parameters for evaluation. Therefore, it is necessary to use the present invention to realize the evaluation of reservoir geological-engineering heterogeneity considering multiple factors, obtain the comprehensive score profile of the whole well section reflecting the potential production, and optimize the best perforation position.

[0117] Substitute the geological parameters of Well N1 and the average engineering parameters of Class C wells into the geological-engineering fuzzy comprehensive evaluation model of Class C wells. The differential evaluation results of reservoir heterogeneity are as Figure 6 shown. The score of the horizontal section of Well N1 is between 42.81 and 74.28. Among them, the score of the well sections at 4030 - 4050 m and 4200 - 4300 m is relatively low and can be regarded as the "low score section", while the score range of the remaining well sections varies greatly and the high score positions are prominent, which are regarded as the "high score section". And Figure 6 shows that the geological parameters of the "low score section" are poor and homogeneous, indicating that the final evaluation results effectively reflect the heterogeneous characteristics of the reservoir and have a good corresponding relationship with the geological parameters.

[0118] Thus, it can be seen that the method proposed by the present invention can clarify the potential correlation relationship between the source rock properties, the fracturability and the reservoir stimulation degree and the production, and realize the differential evaluation of the geological-engineering heterogeneity of shale oil fracturing horizontal wells.

[0119] Although the specific implementation manners of the present invention have been described in detail with reference to the accompanying drawings, it should not be construed as a limitation on the protection scope of this patent. Within the scope described in the claims, various modifications and deformations that can be made by those skilled in the art without creative efforts still fall within the protection scope of this patent.

Claims

1. An evaluation method for heterogeneity differentiation of fractured horizontal wells in unconventional reservoirs, comprising the following steps: 1) Screen historical wells with geological characteristics similar to those of the well to be fractured as sample wells through geological factor clustering, and obtain the geological parameters and engineering parameters of the sample wells; 2) Standardize the geological parameters and engineering parameters of the sample wells, and linearly combine the standardized parameters of each well into principal components through principal component analysis; 3) Determine the membership degree of the principal components to the evaluation levels through the Gaussian distribution membership function, determine the principal component weights through the entropy weight method, compound the principal component membership degree matrix and the weight matrix to obtain the fuzzy comprehensive score of a single well, and establish an evaluation method for heterogeneity differentiation of fractured horizontal wells in unconventional reservoirs; 4) Substitute the geological parameters of the horizontal section of the well to be fractured and the average engineering parameters of the sample wells into the model to obtain the fuzzy comprehensive scoring profile of the horizontal section of the well to be fractured, and realize the evaluation of heterogeneity differentiation of the well to be fractured.

2. The evaluation method for heterogeneity differentiation of fractured horizontal wells in unconventional reservoirs according to claim 1, wherein in step 1), the sample wells are screened through the FCM clustering algorithm, and the FCM clustering algorithm is: Where n is the number of factors in the data set x; α is the number of subsets; μ ij is the membership degree, indicating the similarity degree between the data point x i and the subset j, and the constraint condition is m is the fuzziness parameter; ν j is the class center value of the j-th subset.

3. The evaluation method for heterogeneity differentiation of fractured horizontal wells in unconventional reservoirs according to claim 1, wherein the geological parameters include natural gamma, total organic carbon content, pyrolysis parameters, porosity, permeability, Young's modulus, Poisson's ratio, minimum horizontal principal stress, and horizontal principal stress difference coefficient.

4. The evaluation method for heterogeneity differentiation of fractured horizontal wells in unconventional reservoirs according to claim 1, wherein the engineering parameters include average section length, average cluster spacing, displacement per unit length, sand addition intensity, and liquid usage intensity.

5. The evaluation method for heterogeneity differentiation of fractured horizontal wells in unconventional reservoirs according to claim 1, wherein the standardization equation in step 2) is: In the formula, is the standardized value, x ij is the value of the yield influencing factor; minx ij is the minimum value of the corresponding factor in the sample; maxx ij is the maximum value of the corresponding factor in the sample.

6. The evaluation method for heterogeneity differentiation of fractured horizontal wells in unconventional reservoirs according to claim 1, wherein step 2) further includes the step: Obtain the eigenvalues λ by normalizing the covariance matrix of the dataset 1 ≥ λ 2 ≥ … ≥ λ n ≥ 0 and the corresponding eigenvectors v 1 , v 2 , …, v n , where v j = (v 1j , v 2j , …, v nj ) T , v nj represents the n-th component of the j-th feature, and each principal component is a linear combination of the original factors: where z n is the n-th principal component; is the standardized value of the n-th factor for the n-th well; The cumulative variance contribution rate of the principal components is: where α k is the cumulative variance contribution rate of the first k principal components, dimensionless.

7. The evaluation method for heterogeneity differentiation of fractured horizontal wells in unconventional reservoirs according to claim 1, wherein step 3) further includes: According to the scores assigned to each level, convert the fuzzy comprehensive evaluation result into an intuitive score: where ξ is the fuzzy comprehensive score; f i represents the membership degree of this well to the i-th level; d i is the score; Establish an evaluation matrix based on m wells and k principal components in the sample well system: The entropy value e of each principal component j The determination method is as follows: In the formula, is the proportion of each sample point in this principal component; e j is the entropy value of the j-th principal component, dimensionless; The entropy weight of each principal component is: where w j is the weight of the j-th principal component, 0 ≤ w j ≤ 1 and obtain the principal component weight matrix: W = [w 1 … w j … w k 1×k (12).​

Images