Method for identifying layered boundary based on fuzzy clustering and centimeter-scale water injection test profile

Abstract

The invention discloses a method for identifying a layered boundary based on fuzzy clustering and a centimeter-scale water injection test profile, and belongs to the technical field of underground water science and technology, and the method comprises the following steps: S1, screening centimeter-scale water injection test profile data; s2, performing fuzzy clustering analysis on the water injection pressure profile data; s3, data grouping iterative optimization is carried out; s4, hierarchical combination is carried out according to the identification result, and a hierarchical boundary is obtained; on the basis of centimeter-scale water injection test profile data, a fuzzy clustering analysis method is adopted, a clustering center is defined, and part of member relations of each observation value in each water injection test profile in the clustering process are analyzed, so that a good clustering number is found out, automatic layering of vertical change profile data is realized, and the clustering accuracy is improved. And the uncertainty of the aquifer model caused by inaccurate boundary identification is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of groundwater science, and specifically relates to a method for identifying stratigraphic boundaries based on fuzzy clustering and centimeter-scale injection test profiles. Background Art

[0002] In the field of groundwater science, accurately describing the hydraulic parameters of aquifers is crucial for the correct implementation of flow and solute transport models; the spatial variations of aquifer hydraulic parameters (such as hydraulic conductivity and storage coefficient) exhibit multi-scale characteristics such as randomness and geological stratification; traditional aquifer characterization techniques, such as pumping tests, can obtain spatially averaged estimates of hydraulic conductivity and storage coefficient, but it is difficult to finely present their spatial distributions.

[0003] In order to more finely and quickly capture the spatial distribution of hydraulic conductivity, logging technology has emerged. By conducting centimeter-scale injection tests in the vertical direction in unconsolidated sediments to obtain injection test profiles, the continuous changes in injection pressure can be recorded. This method has the advantages of fast speed and low cost compared with traditional techniques, and can also perform intensive measurements in unconsolidated sedimentary aquifers; centimeter-scale injection test profiles can not only be used to derive the vertical distribution of hydraulic conductivity, but also provide stratification information in the vertical direction of the aquifer; its principle is that the measured injection pressure is related to the grain size and compositional characteristics of the aquifer sediments. Fine-grained materials require higher injection pressures, while coarse-grained unconsolidated materials require lower injection pressures, so it can provide reliable observational data for the permeability changes of different types of sediments.

[0004] However, in previous studies and actual site work, although centimeter-scale injection test profiles have a certain role in inferring geological layers, there are major problems in estimating and identifying geological layer boundaries: the currently commonly used subjective judgment methods are greatly affected by human factors and have uncertainties, and cannot ensure the objective, stable, and effective identification of the stratigraphic boundaries of injection test profiles; if only a fixed threshold is used to determine the boundaries of all profiles, in the case of lack of preliminary data on the actual site stratigraphic conditions, it may make it more difficult to establish a refined aquifer model, resulting in a large uncertainty in the aquifer model and affecting the research and application of groundwater-related problems. Summary of the Invention

[0005] Problems to be Solved

[0006] In view of the problems raised in the existing background art, the present invention provides a method for identifying stratigraphic boundaries based on fuzzy clustering and centimeter-scale injection test profiles.

[0007] Technical Solution

[0008] To solve the above problems, the present invention adopts the following technical solutions.

[0009] A method for identifying layered boundaries based on fuzzy clustering and centimeter-scale water injection test profiles, comprising the following steps:

[0010] Step S1, screening of centimeter-scale water injection test profile data: eliminating sample data that exceeds the reasonable range and cannot be reasonably explained during the continuous water injection test;

[0011] Step S2, fuzzy clustering analysis of water injection pressure profile data: processing the water injection pressure profile data using fuzzy clustering analysis;

[0012] Step S3, data grouping iterative optimization: for a specific centimeter-scale water injection test data profile, adjust the clustering group N value, and calculate the FPI value, MPE value, and J value corresponding to each N value, and make a comparison and judgment after calculation;

[0013] Step S4, obtaining the layered boundary according to the identification result stratification combination: after clustering the water injection pressure profile data using fuzzy clustering analysis, classify the clustering of typical values in the water injection pressure profile data according to the regional boundary information provided by the fuzzy clustering analysis clustering, realize the stratification combination of the identification result under the optimal grouping number N value, and define the layered boundary.

[0014] Preferably, when screening and eliminating the experimental profile data in step S1, the direct elimination method is used to directly eliminate the sample data that exceeds the reasonable range and cannot be reasonably explained.

[0015] Furthermore, when using fuzzy clustering analysis to process the profile data in step S2, the specific steps are as follows:

[0016] When there is a multivariate data set with n data containing p variables, fuzzy clustering analysis finds the optimal number of clusters in the multivariate data set, defines the cluster centers, and calculates the partial membership of each observation in the cluster under the conditions that the clusters are non-exclusive, jointly exhaustive, and non-empty, and realizes clustering by minimizing the objective function. The formula of the objective function is:

[0017]

[0018] C is the cluster center matrix, M is the partial membership matrix, the total membership value of each data is 1, μ ij is the partial membership of the i-th data in the j-th cluster, φ is the fuzzy index (φ>1), d ij is the distance between the i-th data and the j-th cluster center.

[0019] Furthermore, in the fuzzy clustering analysis, the fuzzy performance index and the modified partition entropy are used as evaluation indicators for the clustering performance of the fuzzy clustering analysis. The calculation formula of the fuzzy performance index is:

[0020]

[0021] The calculation formula for the corrected partition entropy is as follows:

[0022]

[0023] F represents the partition coefficient, and H represents the entropy function.

[0024] Furthermore, in the calculation formulas for the fuzzy performance index and the corrected partition entropy, the expressions of F and H are as follows:

[0025]

[0026] μ ij is the partial membership of the i-th data in the j-th cluster, and φ is the fuzzy index.

[0027] Still further, in step S3, data grouping iterative optimization is performed, and the range of the adjusted clustering group number N value is N = 2 - 100.

[0028] Beneficial effects

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0030] (1) With the aid of fuzzy clustering analysis, the present invention constructs a fuzzy matrix based on the data's own attributes to determine the clustering relationship, and objectively clusters the injection pressure profile data; in specific implementation, the optimal number of groups is determined through rigorous calculation and analysis, thereby accurately identifying the stratification boundary, avoiding subjective randomness, and providing more accurate formation stratification information for groundwater research;

[0031] By accurately dividing the stratification boundary and providing a reliable stratification result, the error in model construction is significantly reduced; for example, the vertical stratification information obtained in the embodiment lays a solid foundation for constructing a high-precision aquifer model, enhances the accuracy of the model's simulation of water flow movement and solute transport, and improves the reliability of research on groundwater-related issues.

[0032] In the data processing stage of the present invention, the direct elimination method is used to screen data, effectively removing outliers and ensuring the reliability of the analysis data; in fuzzy clustering analysis, the fuzzy performance index (FPI) and the corrected partition entropy (MPE) are introduced to evaluate the clustering effect, which can evaluate the clustering quality from multiple dimensions; in the process of adjusting the clustering group N value, the FPI, MPE, and J values are comprehensively calculated and compared, and the optimal clustering number can be accurately found, realizing in-depth mining and efficient analysis of data, and giving full play to the value of centimeter-scale injection test profile data. Brief description of the drawings

[0033] To more clearly illustrate the technical solutions in the embodiments or exemplifications of the present application, the following will briefly introduce the accompanying drawings required for use in the embodiments or exemplifications. Obviously, the accompanying drawings in the following description are only some embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the ones shown in these drawings.

[0034] Figure 1 It is the system flow chart of the present invention;

[0035] Figure 2 It is the pressure profile diagram of the centimeter-scale water injection test of the present invention;

[0036] Figure 3 It is the schematic diagram of the clustering result obtained by using the FCM algorithm when the number of groups of the water injection pressure profile in the embodiment of the present invention increases from 2 to 18;

[0037] Figure 4 It is the schematic diagram of the change of the fuzzy performance index and the corrected partition entropy value when the number of groups of the water injection pressure profile in the embodiment of the present invention increases from 2 to 18;

[0038] Figure 5 It is the schematic diagram of the optimal grouping result identified in the embodiment of the present invention and the acquisition of the stratification boundary after hierarchical combination. Specific embodiments

[0039] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Usually, the components of the embodiments of the present application described and shown in the accompanying drawings here can be arranged and designed in various different configurations.

[0040] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the claimed present application, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0041] Embodiment 1

[0042] As Figure 1 shown, a method for identifying the stratification boundary based on fuzzy clustering and centimeter-scale water injection test profiles includes the following steps:

[0043] Step S1. Screening of centimeter-scale water injection test profile data: excluding sample data that exceeds the reasonable range and cannot be reasonably explained during the continuous water injection test;

[0044] Step S2. Fuzzy clustering analysis of water injection pressure profile data: processing the water injection pressure profile data using fuzzy clustering analysis;

[0045] Step S3. Data grouping iterative optimization: for a specific centimeter-scale water injection test data profile, adjust the clustering group N value, calculate the FPI value, MPE value, and J value corresponding to each N value, and make a comparison and judgment after the calculation;

[0046] Refer to Figure 2 , it should be noted that determine whether the N value is lower than the FPI value, MPE value, and J value when the number of groups is N - 1. If it is lower than the corresponding values when N - 1, the number of groups is N + 1, and repeat this calculation; repeat the calculation until N reaches Nmax

[0047] Step S4. Stratified combination according to the recognition result to obtain the stratification boundary: after clustering the water injection pressure profile data using fuzzy clustering analysis, classify the clustering of typical values in the water injection pressure profile data according to the regional boundary information provided by the fuzzy clustering analysis clustering, realize the stratified combination of the recognition result under the optimal number of groups N value, and define the stratification boundary.

[0048] When screening and excluding the experimental profile data in Step S1, the direct elimination method is used to directly exclude sample data that exceeds the reasonable range and cannot be reasonably explained.

[0049] When using fuzzy clustering analysis to process the profile data in Step S2, the specific steps are as follows:

[0050] When there is a multivariate data set with n data containing p variables, fuzzy clustering analysis finds the optimal number of clusters in the multivariate data set, defines the cluster centers, and calculates the partial membership of each observation in the cluster under the conditions that the clusters are non-exclusive, jointly exhaustive, and non-empty, and realizes clustering by minimizing the objective function. The objective function formula is:

[0051]

[0052] C is the cluster center matrix, M is the partial membership matrix, the total membership value of each data is 1, μ ij is the partial membership of the i-th data in the j-th cluster, φ is the fuzzy index (φ > 1), d ij is the distance between the i-th data and the j-th cluster center.

[0053] In fuzzy clustering analysis, the fuzzy performance index and the modified partition entropy are used as evaluation indicators for the clustering performance of fuzzy clustering analysis. The calculation formula of the fuzzy performance index is as follows:

[0054]

[0055] The calculation formula of the modified partition entropy is as follows:

[0056]

[0057] F represents the partition coefficient, H represents the entropy function. The value of FPI ranges between 0 and 1. FPI = 0 indicates perfect clustering of data and no overlap between clusters, and FPI = 1 indicates the worst clustering of data and all clusters are completely overlapped.

[0058] MPE defines the certainty of FCM clustering, and MPE = 0 corresponds to the maximum certainty.

[0059] In the calculation formulas of the fuzzy performance index and the modified partition entropy, the expressions of F and H are as follows:

[0060]

[0061] μ ij is the partial membership of the i-th data in the j-th cluster, and φ is the fuzzy index.

[0062] In step S3, data grouping iterative optimization is performed, and the range of the adjusted clustering group number N value is N = 2 - 100.

[0063] Referring to the above steps, in specific implementation, the method is as follows:

[0064] Screening of centimeter-scale water injection test profile data: When conducting centimeter-scale water injection tests, high-precision pressure sensors are used to record the data of the water injection pressure varying with time and depth.

[0065] During the data acquisition process, it is found that some data are abnormal; for example, at a certain depth point, the water injection pressure suddenly increases, far exceeding the normal water injection pressure range of this area, and after checking the equipment and the surrounding environment, no reasonable reason for the abnormal increase in pressure is found; at this time, the direct elimination method is used to directly eliminate these sample data that exceed the reasonable range and cannot be reasonably explained from the data set to ensure the reliability of the subsequent analysis data.

[0066] Fuzzy clustering analysis of water injection pressure profile data: After data screening, a multivariate data set of 50 data samples including 3 variables of water injection pressure, water injection time, and water injection depth is obtained.

[0067] Set the fuzzy index and process these data using fuzzy clustering analysis technology; during the clustering process, the fuzzy clustering algorithm searches for the optimal number of clusters in the multivariate dataset, and finally determines that the number of clusters is 4, and defines the corresponding cluster centers; according to the conditions of non-exclusive, jointly exhaustive, and non-empty clustering, calculate the partial membership of each observation in the cluster, and through continuous iterative update, minimize the objective function to achieve effective clustering of the data.

[0068] Iterative optimization of data grouping: For the data profile of the centimeter-scale water injection test, adjust the clustering group N value in the range of 2 - 100.

[0069] In specific implementation, starting from N = 2, calculate the values of the fuzzy performance index (FPI), modified partition entropy (MPE), and objective function (J) corresponding to each N value in turn.

[0070] When N = 2, the calculated FPI value is 0.6, the MPE value is 0.4, and the J value is 15;

[0071] When N = 3, the FPI value is 0.5, the MPE value is 0.35, and the J value is 12.

[0072] After each calculation, compare and judge these values. As the N value increases, the J value continues to decrease, and the FPI value and MPE value also show a downward trend;

[0073] In specific implementation, after multiple calculations and comparisons, it is found that when N = 6, both the FPI value and the MPE value reach relatively low levels, and the clustering result is optimal at this time.

[0074] Stratified combination according to the recognition result to obtain the stratification boundary: After clustering the water injection pressure profile data using fuzzy clustering analysis, according to the regional boundary information provided by the clustering, classify the clustering of typical values in the water injection pressure profile data.

[0075] For example, cluster the data with relatively low water injection pressure and similar change trends into the same category; in the case of the optimal number of groups N value (N = 6), the stratified combination of the recognition results is realized, and the stratification boundary is clearly defined. In this way, more accurate stratification information of the aquifer in this area in the vertical direction is obtained, providing a reliable basis for establishing a refined aquifer model.

[0076] Compared with the previous method of determining the stratification boundary by subjective judgment or fixed threshold, the stratification result provided by this method is more objective and stable, effectively reducing the uncertainty of the aquifer model.

[0077] In summary, the present invention utilizes fuzzy clustering analysis to construct a fuzzy matrix based on the inherent attributes of the data to determine the clustering relationship, objectively clustering the water injection pressure profile data. During the specific implementation, the optimal number of groups is determined through rigorous calculations and analyses, thereby accurately identifying the stratification boundaries, avoiding subjective randomness, and providing more accurate formation stratification information for groundwater research.

[0078] By accurately dividing the stratification boundaries and providing reliable stratification results, the error in model construction is significantly reduced. For example, the vertical stratification information obtained in the embodiment lays a solid foundation for constructing a high-precision aquifer model, enhancing the accuracy of the model's simulation of water flow movement and solute transport, and improving the reliability of groundwater-related problem research.

[0079] In the data processing stage of the present invention, the direct elimination method is adopted to screen the data, effectively removing outliers and ensuring the reliability of the analysis data. During the fuzzy clustering analysis, the fuzzy performance index (FPI) and the modified partition entropy (MPE) are introduced to evaluate the clustering effect, enabling the evaluation of the clustering quality from multiple dimensions. During the process of adjusting the clustering group N value, the FPI, MPE, and J values are comprehensively calculated and compared to accurately find the optimal clustering number, realizing the in-depth mining and efficient analysis of the data, and fully exerting the value of the centimeter-scale water injection test profile data.

[0080] The above-described embodiments only represent the preferred implementation modes of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the present invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several variations, improvements, and substitutions can be made, and these all fall within the protection scope of the present invention.

Claims

1. A method for identifying stratification boundaries based on fuzzy clustering and centimeter-scale water injection test profiles, characterized in that, It includes the following steps: Step S1, screening of centimeter-scale water injection test profile data: eliminating sample data that exceeds the reasonable range and cannot be reasonably explained during the continuous water injection test; Step S2, fuzzy clustering analysis of water injection pressure profile data: processing the water injection pressure profile data using fuzzy clustering analysis; Step S3, data grouping iterative optimization: for a specific centimeter-scale water injection test data profile, adjusting the clustering group N value, calculating the FPI value, MPE value, and J value corresponding to each N value, and making a comparison and judgment after the calculation; Step S4, obtaining the stratification boundary according to the recognition result stratification combination: after clustering the water injection pressure profile data using fuzzy clustering analysis, according to the regional boundary information provided by the fuzzy clustering analysis clustering, classifying the clustering of typical values in the water injection pressure profile data, realizing the stratification combination of the recognition result under the optimal case of the grouping number N value, and defining the stratification boundary.

2. A method for identifying layered boundaries based on fuzzy clustering and centimeter-scale water injection test profiles according to claim 1, characterized in that: When screening and eliminating the experimental profile data in Step S1, the direct elimination method is adopted to directly eliminate the sample data that exceeds the reasonable range and cannot be reasonably explained.

3. A method for identifying a layered boundary based on fuzzy clustering and a centimeter-scale water injection test profile according to claim 1, characterized in that: When using fuzzy clustering analysis to process the profile data in Step S2, the specific steps are as follows: When there is a multivariate data set of n data containing p variables, fuzzy clustering analysis finds the optimal number of clusters in the multivariate data set, defines the cluster centers, and calculates the partial membership of each observation in the cluster under the conditions that the clusters are non-exclusive, jointly exhaustive, and non-empty, and realizes clustering by minimizing the objective function. The formula of the objective function is: C is the cluster center matrix, M is the partial membership matrix, and the total membership value of each data sums to 1. μ ij is the partial membership of the i-th data in the j-th cluster, φ is the fuzzy index (φ > 1), and d ij＝ is the distance of the i-th data from the j-th cluster center.

4. A method for identifying stratification boundaries based on fuzzy clustering and centimeter-scale water injection test profiles according to claim 3, characterized in that: In the fuzzy clustering analysis, the fuzzy performance index and the modified partition entropy are used as the evaluation indexes of the clustering performance of the fuzzy clustering analysis. The calculation formula of the fuzzy performance index is: The calculation formula of the modified partition entropy is: F represents the partition coefficient, and H represents the entropy function.

5. A method for identifying layered boundaries based on fuzzy clustering and centimeter-scale water injection test profiles according to claim 4, characterized in that: In the calculation formulas of the fuzzy performance index and the modified partition entropy, the expressions of F and H are as follows: μ ij is the partial membership of the i-th data in the j-th cluster, and φ is the fuzzy exponent.

6. A method for identifying layered boundaries based on fuzzy clustering and centimeter-scale water injection test profiles according to claim 5, characterized in that: In Step S3 for data grouping iterative optimization, the range of adjusting the clustering group number N value is N = 2 - 100.

Images