Oil-water well control property determination method and device based on fuzzy analytic hierarchy process

Abstract

The application provides a fuzzy analytic hierarchy process-based oil-water well lifting-control property determination method and device, and relates to the oil and gas field development technical field.The method comprises the following steps: determining multiple indexes affecting oil well end-control decision, including daily oil production, daily liquid production, daily water production, water cut, single-layer oil production and oil production intensity, etc.The analytic hierarchy process is used to weight each index, the weight is determined by introducing statistical analysis, so that the weight calculation is more objective and accurate.Each index value is mapped to the membership degree of oil well end-control or water well end-lifting by using a fuzzy membership function.The weighted membership degrees of all indexes are comprehensively considered, the comprehensive evaluation value of a single well in the "oil well end-control" and "water well end-lifting" two categories is calculated, and the lifting-control property of the well is determined accordingly.The method can comprehensively consider multiple hierarchical information and uncertain factors, improve the accuracy and scientificity of the lifting-liquid control-water decision, and is suitable for the comprehensive management and development optimization of high-water-cut oil fields.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of oil and gas field development, and particularly relates to a method and equipment for determining lifting and controlling properties of oil and water wells based on fuzzy analytic hierarchy process. BACKGROUND

[0002] In the middle and late development stage of an oilfield, the injection-production system tends to be unbalanced, the water cut of oil wells increases significantly, and the production gradually decreases, which becomes the main factor restricting the increase of recovery rate. Therefore, the "lifting and controlling" measures are usually taken in engineering: the water control and oil stabilization are implemented for oil wells, and the injection increase and pressure stabilization are implemented for injection wells. How to accurately identify the "lifting" or "controlling" properties of oil and water wells is a core problem for formulating a scientific development plan and a dynamic control strategy. However, the existing determination methods mainly depend on experience or a single parameter threshold, and the results are subjective and have poor repeatability, so it is difficult to form a unified standard under complex conditions of multiple wells and multiple layers.

[0003] The prior art extends the fuzzy comprehensive evaluation to the injection-production well connectivity analysis field, determines the weights of various factors by using an entropy weight method, and performs fuzzy comprehensive evaluation on the dominant channels of various injection-production directions according to the maximum membership degree principle. The model has certain application value in well pattern system analysis, but mainly focuses on the injection-production relationship, rather than the "liquid lifting / water control" properties of single wells.

[0004] In addition, the patent CN117786922A relates to a method and system for determining a reasonable water drive lower limit of a horizontal well in an ultra-low permeability oil reservoir, and a fuzzy analytic hierarchy process is used in the evaluation of the development effect of the oil reservoir to comprehensively identify the development types (stable type, decreasing type, and water breakthrough type) of the well, and has high rationality of the model structure. However, the application scenario of the patent mainly faces the development effect classification, and the evaluation indexes are relatively macro, such as the production capacity decline rate and the water cut increase, and the analysis is not carried out for the "liquid lifting / water control" sub-attribute. At the same time, the judgment matrix is still based on expert scoring, and the weight system lacks objective data basis.

[0005] Overall, the existing research has made certain progress in the identification of the state of oil and water wells, but still has common deficiencies: (1) the weight determination depends on subjective judgment and lacks data support; (2) the membership degree function setting is highly experienced and it is difficult to reflect the distribution characteristics of the indexes; (3) the model consistency and robustness test is missing; and (4) it is difficult to adapt to the dynamic changes of different oil reservoirs and development stages. SUMMARY

[0006] The purpose of the present application is to provide a method for determining the lifting and controlling properties of oil and water wells based on fuzzy analytic hierarchy process, in order to solve the problem that the existing oil and water well state identification method cannot accurately identify.

[0007] The above-mentioned purpose of the present application is achieved by the following technical solutions:

[0008] S1: Obtain production dynamic index data and pre-process, construct sample data set;

[0009] S2: According to the actual liquid extraction or water control attribute, the sample data set is divided into two categories, and single factor variance analysis is carried out respectively to obtain a weight coefficient vector;

[0010] S3: Based on the weight coefficient vector, the fuzzy membership function of the production dynamic index is constructed for the oil well end liquid extraction and the oil well end water control, and the liquid extraction membership degree and the water control membership degree are obtained;

[0011] S4: Using the difference comparison fuzzy decision rule, combining the fuzzy membership function, the extraction and control attributes of the oil and water well are classified.

[0012] Optionally, step S1 comprises:

[0013] The daily production dynamic index data is extracted from the oil well and water injection well production database of the target oilfield block and pre-processed to construct a sample data set; the production dynamic index data includes daily oil production, daily liquid production, daily water production, water cut, single layer oil production and oil production intensity;

[0014] The pre-processing step comprises: performing abnormal correction and scale transformation on the production dynamic index data;

[0015] The abnormal correction of the production dynamic index data comprises: using the quantile truncation method to process the abnormal value of the production dynamic index data, setting the predetermined lower quantile point Ql and the upper quantile point Qu, and adjusting the abnormal value exceeding the [Ql, Qu] interval to the corresponding boundary value;

[0016] The scale transformation of the production dynamic index data comprises:

[0017] The daily oil production, daily liquid production and oil production intensity are first logarithmically transformed and then square rooted;

[0018] The daily water production and single layer oil production are logarithmically transformed;

[0019] The water cut is processed by inverse complement and logarithmic transformation.

[0020] Optionally, step S2 comprises:

[0021] S21: Calculate the single factor variance analysis effect of each production dynamic index And convert to get dimensionless f value; f value is used to represent the discrimination strength of production dynamic index;

[0022] S22: Normalize all f values to generate an initial weight vector;

[0023] S23: According to the initial weight vector, combined with the discriminant ratio of each index, a judgment matrix of the analytic hierarchy process is constructed, and the relative importance of any two indexes in the judgment matrix is determined by the normalized discriminant ratio, and is mapped to the 1-9 scale;

[0024] The maximum eigenvalue and consistency ratio CR of the judgment matrix are calculated, and when CR<0.1, the consistency of the judgment matrix is acceptable; if CR exceeds 0.1, the index weight ratio is adjusted or the highly correlated indexes are removed, and then recalculated until the consistency requirement is met;

[0025] S24: The weight coefficient vector W of each production dynamic index is calculated by using the geometric mean method combined with the judgment matrix, and the weight coefficient vector is used to represent the relative influence of each index in the liquid extraction / water control judgment.

[0026] Optionally, step S3 comprises:

[0027] S31: A Gaussian function is used to determine the fuzzy membership function; the fuzzy membership function includes: a liquid extraction membership function and a water control membership function; the sample data set includes: a liquid extraction well sample and a water control well sample;

[0028] S32: For each production dynamic index, the fuzzy grade interval is divided according to the value distribution of the liquid extraction well sample and the water control well sample, and the lower quartile q1, the median q2 and the upper quartile q3 of the index in the two types of well samples are obtained;

[0029] The median q2 is selected as the center value c of the Gaussian membership function, and a preset proportion of the interquartile range is selected as the estimation of the standard deviation σ;

[0030] The expression form of the Gaussian membership function is:

[0031]

[0032] Wherein is the normalized value of the index, is the center parameter of the index for the target category, is the width parameter;

[0033] S33: The Gaussian membership function is used to map the continuous numerical production dynamic index to the membership degree value of the liquid extraction attribute and the water control attribute.

[0034] Optionally, step S33 comprises:

[0035] For each well to be judged, the production dynamic indexes of the well are substituted into the corresponding liquid extraction membership function and water control membership function to obtain the fuzzy membership degree matrix of the well relative to the liquid extraction attribute and the water control attribute;

[0036] Based on the weight coefficient vector, the membership matrix is ​​weighted and summed to calculate the membership degree of the well belonging to the fluid extraction attribute. And membership degree belonging to water control attributes .

[0037] Optionally, step S4 includes:

[0038] Determine the membership degree of each well's fluid extraction. water control membership degree Comparative analysis was conducted, and based on pre-set fuzzy judgment rules, the lifting and control properties of oil and water wells were finally determined, yielding the judgment results, specifically including:

[0039] Calculate the membership difference And judge based on the preset experience threshold:

[0040] when Greater than the positive threshold At that time, the well was identified as a fluid extraction well;

[0041] when Less than the negative threshold At that time, the well was identified as a water control well;

[0042] when The absolute value is lower than the preset threshold If the condition is met, the well is considered stable.

[0043] Optionally, the method further includes: batch processing and model adaptive adjustment mechanism and periodic correction;

[0044] The batch processing and model adaptive adjustment mechanism specifically includes: performing batch judgment analysis on multiple oil and water wells; using a pre-set list of well numbers and target date batches, the system automatically extracts the latest valid production data snapshot for each well for the corresponding time period; and performs data preprocessing, weight calculation, membership evaluation and control judgment in sequence according to steps S1–S4; and outputs the membership results of fluid extraction and water control for each well and the recommended control category in batches.

[0045] Periodic calibration includes: periodically calibrating the entire algorithm model using real-time updated production dynamic data, and adjusting the weight coefficient vector W and the preset empirical threshold δ based on changes in well conditions.

[0046] An electronic device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to enable the electronic device to perform a method for determining the controllability of oil and water wells based on fuzzy hierarchical analysis.

[0047] A computer readable storage medium stores instructions that, when executed, perform an oil-water well lifting control property determination method based on fuzzy analytic hierarchy process.

[0048] The technical scheme provided by the application has the beneficial effects that:

[0049] By introducing systematic data collection, cleaning and standardized processing to multiple dynamic production indexes of oil-water wells, the objectivity and repeatability of the lifting control determination process are realized; combined with variance analysis, effect evaluation and analytic hierarchy process to calculate the index weight, the contribution of each production parameter in the determination is quantified, and the influence of artificial experience subjectivity on the result is significantly reduced; by establishing a Gaussian fuzzy membership function and a fuzzy comprehensive evaluation model, continuous membership determination of well layer liquid lifting and water control properties is realized, effectively processing the uncertainty and boundary fuzzy situation between indexes, improving the determination accuracy and adaptability; the method supports multi-well batch analysis and dynamic feedback correction, so that the model has self-correction ability and continuous optimization performance; finally, the application can realize high automation and intelligentization of lifting control determination, not only improving the determination efficiency, but also providing a scientific basis for oilfield injection-production optimization, well pattern adjustment and production decision-making, which has important application value in actual oilfield production management. BRIEF DESCRIPTION OF DRAWINGS

[0050] The application will be further described below in combination with the drawings and examples, and the drawings are as follows:

[0051] Figure 1 is a step diagram in the embodiment of the application;

[0052] Figure 2 is a structure diagram of the application in the embodiment of the application;

[0053] Figure 3 is a data distribution schematic diagram in the embodiment of the application;

[0054] Figure 4 is a single factor variance analysis effect diagram in the embodiment of the application;

[0055] Figure 5 is a judgment matrix schematic diagram in the embodiment of the application;

[0056] Figure 6 is a membership function schematic diagram in the embodiment of the application;

[0057] Figure 7 is a lifting control effect well position schematic diagram in the embodiment of the application;

[0058] Figure 8 is an electronic device structure schematic diagram in the embodiment of the application. DETAILED DESCRIPTION

[0059] In order to have a clearer understanding of the technical features, objectives and effects of the present application, the specific embodiments of the present application will be described in detail with reference to the drawings.

[0060] The embodiment of the present application provides a fuzzy analytic hierarchy process-based oil-water well lifting control property determination method.

[0061] Please refer to Figure 1 , Figure 1 is a step diagram of a fuzzy analytic hierarchy process-based oil-water well lifting control property determination method in the embodiment of the present application, comprising:

[0062] S1: acquiring production dynamic index data and performing preprocessing to construct a sample data set;

[0063] S2: according to actual lifting or water control properties, the sample data set is divided into two categories, and single-factor variance analysis is respectively performed to obtain a weight coefficient vector;

[0064] S3: based on the weight coefficient vector, fuzzy membership functions of two properties of oil well end lifting and oil well end water control are respectively constructed for the production dynamic index, and lifting membership degree and water control membership degree are obtained;

[0065] S4: adopting a difference comparison fuzzy decision rule, the lifting and control properties of the oil-water well are classified in combination with the fuzzy membership functions.

[0066] The present application provides an embodiment as follows, which belongs to oilfield development production dynamic analysis and adjustment technology, and is used for guiding the decision of oil well yield increase lifting or water injection well water control allocation. Figure 2 is a structural diagram of the present application in the embodiment of the present application.

[0067] As an embodiment, a fuzzy analytic hierarchy process (Fuzzy AHP)-based oil-water well lifting control property comprehensive determination method is used for judging the lifting water control category of the wellbore in the oilfield development process.

[0068] Step S1 comprises:

[0069] The daily production dynamic index data is extracted from the oil well and water injection well production database of the target oilfield block and is preprocessed to construct a sample data set; the production dynamic index data comprises daily oil production, daily liquid production, daily water production, water cut, single-layer oil production and oil production intensity;

[0070] The preprocessing step comprises: performing abnormal correction and scale transformation on the production dynamic index data;

[0071] The abnormal correction of the production dynamic index data specifically includes: using a quantile truncation method to process the abnormal values of the production dynamic index data, setting a predetermined lower quantile point Ql and an upper quantile point Qu, and adjusting the abnormal values exceeding the interval [Ql, Qu] to the corresponding boundary values;

[0072] In one specific embodiment of the present application, the abnormal values exceeding the interval [Ql, Qu] are adjusted to the corresponding boundary values to reduce the influence of extreme values on statistical analysis; and a corresponding transformation strategy is selected according to the data distribution characteristics of different indexes. Generally, Ql = 0.01 quantile and Qu = 0.99 quantile are taken to balance fidelity and robustness.

[0073] The scale transformation of the production dynamic index data specifically includes:

[0074] The daily oil production, daily fluid production and oil production intensity are first logarithmically transformed and then square rooted;

[0075] The daily water production and single-layer oil production are logarithmically transformed;

[0076] The water cut is processed by reverse complementation and logarithmic transformation.

[0077] In one specific embodiment of the present application, the daily oil production, daily fluid production and oil production intensity are first logarithmically transformed and then square rooted to weaken the right-skewed distribution; the daily water production and single-layer oil production are logarithmically transformed to compress the value span; the water cut is processed by “reverse complementation + logarithm” transformation to make it positively correlated with water control effect; after the above transformation, linear dimensionless is uniformly performed on all indexes, which are mapped to the interval [0.1, 1], so as to obtain a standardized sample data set with balanced distribution and consistent dimension.

[0078] In one specific embodiment of the present application, the collected data is cleaned and standardized: missing or abnormal records are removed, extreme abnormal values are truncated by using a quantile truncation method, values exceeding a preset upper and lower quantile range are limited to boundary values; logarithmic or square root transformation is performed on indexes with skewed distribution (the water cut HS index is processed by value inversion and logarithmic transformation to match the water control attribute direction); then linear normalization is performed on all indexes to map the values to the interval [0.1, 1] to eliminate the dimension difference between different wells.

[0079] Step S2 includes:

[0080] S21: calculating a single-factor variance analysis effect quantity of each production dynamic index and converting to obtain a dimensionless f value; the f value is used to represent the strength of the discrimination of the production dynamic index;

[0081] One embodiment provided by the present application is as follows: for each index The variance difference between the two groups of "lift" and "control" is calculated respectively. The significance between groups is tested by one-way ANOVA, and the core statistical quantity is F value:

[0082]

[0083] Among them: : between-group sum of squares; : within-group sum of squares; k = 2: number of groups; N: total sample size. To measure the actual discriminant power of each index, not just rely on significance test, further calculate the effect size and its dimensionless form:

[0084]

[0085]

[0086] The effect size f can reflect the strength of the relationship between the index and the two types of wells "lift" and "control": among them, <0.10: weak discrimination; 0.10≤ <0.25: moderate discrimination; ≥0.40: significant discrimination.

[0087] S22: normalize all f values to generate an initial weight vector;

[0088] S23: According to the initial weight vector, combined with the discriminant power ratio of each index, the judgment matrix of the analytic hierarchy process is constructed. The relative importance of any two indexes in the judgment matrix is determined by the normalized discriminant power ratio, and is mapped to the 1-9 scale;

[0089] Calculate the maximum eigenvalue and consistency ratio CR of the judgment matrix. When CR <0.1, the consistency of the judgment matrix is acceptable; if CR exceeds 0.1, adjust the index weight ratio or remove highly correlated indexes and recalculate until the consistency requirement is met;

[0090] In one embodiment of the present application, the judgment matrix is constructed and the consistency is tested. According to the relative discrimination ratio of each index, the analytic hierarchy process judgment matrix is constructed The matrix elements are calculated as follows:

[0091]

[0092] Among them, the function converts the continuous ratio to the 1-9 hierarchical scale value (if , take its reciprocal). After obtaining the matrix A, solve the maximum eigenvalue λmax, and calculate the consistency index and consistency ratio.

[0093] wherein the consistency index is CI, and the calculation formula is as follows:

[0094]

[0095] The consistency ratio is CR, and the calculation formula is as follows:

[0096]

[0097] wherein is a random consistency index (according to the order of the matrix table). When , it is considered that the matrix consistency is acceptable; if it exceeds, the index importance needs to be adjusted to restructure the matrix.

[0098] S24: The weight coefficient vector W of each production dynamic index is calculated by using the geometric mean method combined with the judgment matrix, and the weight coefficient vector is used to represent the relative influence of each index in the liquid extraction / water control determination.

[0099] In one specific embodiment of the present application, the sample data set is divided into two categories according to the actual liquid extraction or water control attribute, and single factor variance analysis is performed respectively to evaluate the significant difference of each production dynamic index on well classification, and the effect value of each index is calculated, and the dimensionless f value is converted to quantify the strength of the discrimination of each index to distinguish between "liquid extraction well" and "water control well". The f values of all indexes are normalized to generate an initial weight vector, and an analytic hierarchy process (AHP) judgment matrix is constructed according to the discrimination ratio of each index: the relative importance of any two indexes in the matrix is determined by the normalized discrimination ratio, and is mapped to a 1-9 scale. The maximum eigenvalue and consistency ratio CR of the judgment matrix are calculated, and when CR<0.1, the consistency of the judgment matrix is acceptable; if CR exceeds 0.1, the index weight ratio is adjusted according to the deviation or the highly correlated indexes are removed, and then the calculation is performed again until the consistency requirement is met. Finally, the weight coefficient vector W of each index is calculated by using the geometric mean method, and the weight coefficient vector is used to represent the relative influence of each index in the liquid extraction / water control determination.

[0100] Step S3 comprises:

[0101] S31: A Gaussian function is used to determine the fuzzy membership function; the fuzzy membership function comprises a liquid extraction membership function and a water control membership function; the sample data set comprises a liquid extraction well sample and a water control well sample;

[0102] S32: For each production dynamic index, the fuzzy grade interval is divided according to the value distribution of the liquid extraction well sample and the water control well sample, and the lower quartile q1, the median q2 and the upper quartile q3 of the index in the two types of well samples are obtained;

[0103] The median q2 is selected as the center value c of the Gaussian membership function, and a preset proportion of the interquartile range is used as an estimate of the standard deviation σ.

[0104] The expression of the Gaussian membership function is:

[0105]

[0106] wherein is a standardized value of the index, is a center parameter of the index for the target category, is a width parameter;

[0107] In one specific embodiment of the present application, for each index, the statistical characteristics of the quantile of the index in the two categories are extracted by analyzing the value distribution of the two types of samples of the liquid production well and the water control well, wherein the median is taken as the center value c of the Gaussian membership function, and the width parameter σ is estimated according to the difference between the upper quartile and the lower quartile. By the above method, the fuzzy membership function for the “liquid production” attribute and the “water control” attribute is constructed for each index respectively, realizing the conversion of discrete production data to continuous membership degree, and making the attribution degree of the index value of a single well to the two attributes be quantitatively described.

[0108] In one specific embodiment of the present application, the fuzzy grade division and the statistical parameter extraction, after obtaining the standardized data of each index, in order to represent the influence law of different indexes on the “production” and “control” attributes, it is necessary to establish fuzzy grade intervals according to the sample distribution. For each index xi, based on the statistical distribution of the two types of samples, the 25%, 50%, and 75% quantiles are calculated, which are denoted as q1, q2, and q3 respectively, wherein q2 is taken as the center position (i.e. the mean point) of the fuzzy function, and the fuzzy grade intervals are defined according to this: [q1, q3]: main distribution area of the index; the outer area represents the weak feature part. The selection of the quantiles reflects the typical change range of the index in the “production” and “control” two types of samples, avoiding the subjectivity of manual delimitation in the traditional fuzzy model.

[0109] S33: The Gaussian membership function is used to map the continuous numerical production dynamic index to the membership degree values of the liquid production attribute and the water control attribute.

[0110] In one embodiment of the present application, after the weight determination, each indicator needs to be associated with the "lift control" attribute of the well to establish a continuous membership description. For each indicator, by analyzing its distribution characteristics in different categories of samples, fuzzy grade intervals are divided and statistical parameters are extracted, and then a Gaussian membership function is constructed. This process can convert discrete production data into continuous membership, reflecting the fuzzy relationship of each well or each layer sample under the "liquid lifting" and "water control" categories. Subsequently, the membership matrix is combined with the indicator weight vector to calculate the comprehensive membership, realizing the preliminary lift control judgment of single well or single layer. This step plays a bridge role in the model, both connecting data processing and providing input for fuzzy decision-making.

[0111] Step S33 includes:

[0112] For each well to be determined, the production dynamic indicators of the well are substituted into the corresponding liquid lifting membership function and water control membership function to obtain the fuzzy membership matrix of the well with respect to the liquid lifting attribute and the water control attribute;

[0113] According to the weight coefficient vector, the membership matrix is weighted and summed to calculate the membership of the well belonging to the liquid lifting attribute and the membership of the well belonging to the water control attribute .

[0114] The present application provides an embodiment as follows: for each well or each layer sample, the membership of all indicators under two categories of attributes is calculated respectively to form a fuzzy relationship matrix :

[0115]

[0116] Combined with the indicator weight vector W calculated in the previous step, the comprehensive membership vector of each well under different attributes is calculated .

[0117] Step S4 includes:

[0118] The liquid lifting membership and the water control membership of each well are compared and analyzed, and the lift control properties of the oil and water well are finally determined according to the pre-set fuzzy judgment rule to obtain the judgment result, which specifically includes:

[0119] The membership difference is calculated, and the judgment is made according to the pre-set empirical threshold value:

[0120] When is greater than the positive threshold value , the well is determined as a liquid lifting well;

[0121] When is less than the negative threshold value When the absolute value of Δr is lower than a preset threshold value

[0122] When the absolute value of Δr is lower than a preset threshold value , the well is determined as a stable well.

[0123] In one embodiment of the present application, for the wells whose determination results are close to the threshold boundary (i.e. , the system automatically marks as "manually reviewed wells", prompting engineering and technical personnel to further verify in combination with oil well liquid production curves, water cut changes and on-site monitoring data. The determination results are output in tabular or graphical form, each well is attached with the lifting and water control membership degree score and the corresponding optimization adjustment suggestion scheme, which is used to guide the dynamic regulation and control of the oilfield injection-production system. The threshold , , can be set according to the experience of reservoir type and development stage and can be adjusted appropriately to adapt to the determination requirements in different scenarios. When the Δr of any oil well is close to the threshold boundary, the system marks the well as to be manually reviewed to remind further confirmation of its lifting and control properties in combination with on-site actual production data, thereby ensuring the reliability of the decision.

[0124] In one embodiment of the present application, after the comprehensive membership degree vector is completed, the lifting and control properties of each well are finally determined by comparison and analysis with a preset determination rule library. In the rule making, historical production characteristics and on-site experience are fully considered, so that the model can cope with boundary or fuzzy situations, such as outputting "stable well" or prompting manual review. In order to improve the adaptability of the model, real-time production data can be used for periodic feedback and weight adjustment, so that the lifting and control strategy has the ability of dynamic self-correction, thereby continuously optimizing the well pattern management and injection-production scheme.

[0125] The method further comprises a batch processing and model adaptive adjustment mechanism and a periodic correction;

[0126] The batch processing and model adaptive adjustment mechanism specifically comprises: batch determination and analysis of multiple oil and water wells, through a preset well number list and target date batch, the system automatically extracts the latest effective production data snapshot of each well for the corresponding period, and sequentially performs data preprocessing, weight calculation, membership evaluation and lifting and control determination according to the steps S1-S4, and batch outputs the lifting and water control membership degree results and recommended regulation categories of each well;

[0127] The periodic correction comprises: using real-time updated production dynamic data to periodically correct the entire algorithm model, and according to the well condition changes, the weight coefficient vector W and the preset experience threshold δ are implemented feedback adjustment.

[0128] This application provides an embodiment as follows, taking historical production data from 20 oil wells in a certain region as an example. The specific implementation scheme of the present invention is as follows:

[0129] Step 110: Data Preparation and Sample Construction. Obtain historical production data for the target block, organized by well number and date, including at least: daily oil production (RCYL1), daily fluid production (RCYL), daily water production (RCSL), water cut (HS), single-layer oil production (DCYL), oil recovery intensity (CYQD), and historical control tags (DCXZ, if applicable). The data source is a .csv / .xlsx file exported from the production database (Table 1 is a data illustration). First, perform field validation and quality screening: remove records with key indicators of 0 or missing data, and remove unreasonable values ​​(such as negative production, HS>100%). Then, implement a consistent distribution correction strategy: for RCYL1 and RCYL, first trim the tail (1% or 0.5% upper quantile), then take the logarithm and square root; for RCSL, take the logarithm and square root; for CYQD, trim the tail and take the logarithm; for DCYL, take the logarithm; for HS, perform "reverse + logarithmic" processing to match the water control direction. All features are then linearly normalized to [0.1, 1] to ensure comparability across wells. Figure 3 The transformed distribution is given. The samples are finally stored at the well-to-day granularity, serving as a unified input for subsequent statistical discrimination and fuzzy modeling.

[0130] Table 1

[0131]

[0132] Step 120: Discriminative power analysis and AHP weight determination. Divide the samples into two categories, "Improvement" and "Control," based on historical or engineering judgments (if no labels are available, pre- and post-implementation curves or engineering records can be used for initial labeling). Based on this, perform one-way ANOVA on each indicator, outputting F and P, and calculate the effect size η² and Cohen's f to quantify the actual ability to "distinguish between improvement and control." Figure 4 (See the effect diagram). The f-values ​​of each indicator are normalized to a scoring vector S. Using Si / Sj as the ratio, an AHP judgment matrix is ​​constructed through a discrete mapping function (1–9 scale). Figure 5 (This is a matrix illustration). Then, eigenvalue analysis and consistency checks (CI, CR) are performed. The matrix is ​​valid when CR < 0.1; otherwise, the scaling is adjusted or highly correlated weak indicators are removed until the matrix passes. Finally, a stable weight vector W is obtained using the geometric mean method (for example, in this embodiment, HS and RCSL have higher weights, while RCYL and DCYL have relatively lower weights), and this vector is solidified into a unified coefficient set for subsequent fuzzy comprehensive analysis.

[0133] Step 130: Membership function setting and fuzzy comprehensive evaluation, based on the distribution statistics of the "lift / control" two types of samples, 25%, 50%, and 75% quantiles of the six indicators are extracted as the center and width reference of the Gaussian type membership function (q2 is the center, q3 q1 is converted to σ), forming two function families of "well end lift / well end control" Figure 6 ). For any well, the effective small layer on the target day (or the nearest available day), the "lift / control" membership degrees are calculated for each indicator, and assembled into a 6x2 fuzzy relationship matrix M. Then, the weight W obtained in step 120 is weighted to obtain the comprehensive membership degree [Ucontrol, Ulift] of the small layer. The well-level judgment is formed according to the established aggregation strategy: by default, the small layer average is taken; if the block is mainly contributed by thick layers, the DZNET (effective thickness) weighted average is enabled to be closer to the field. The threshold rule is simple and clear: Ucontrol> Ulift is judged as "well end control", and vice versa; if the difference between the two is lower than the engineering set sensitivity threshold, it is marked as "neutral / stable well" for manual review Figure 7 to show the well location and judgment distribution.

[0134] Step 140: Batch reasoning, consistency evaluation and result output, a batch processing list (well number-target date) is established, and the data snapshot of "the current day and the nearest available day after it; if not, the nearest historical day" is automatically selected for each well, which is transformed and normalized according to the same caliber in step 110, and then connected to the membership degree calculation and weighted judgment in step 130. The system outputs two types of results under the same task: one is "single well details", which records the two types of membership degrees, weight weighted results and final small layer categories of each small layer; the other is "well-level aggregation", which includes well-level comprehensive membership degree, final category and use date. If the true category (or the result after field measures) is provided, the accuracy, precision, recall and F1 are automatically calculated, and the confusion matrix and performance bar chart are generated for stage acceptance. All results are exported in the form of tables and drawings for easy traceability Figure 5 、 Figure 6 、 Figure 7 The typical output style is shown in the figure). When the block enters a new stage or the injection-production system changes greatly, it is recommended to update the quantile parameters and weight W to keep consistent with the field.

[0135] The application also discloses an electronic device. Referring to Figure 8 , Figure 8 is a structural schematic diagram of an electronic device disclosed by the embodiment of the application. The electronic device 500 can include at least one processor 501, at least one network interface 504, a user interface 503, a memory 505, and at least one communication bus 502.

[0136] The communication bus 502 is used to realize the connection and communication between the components.

[0137] The user interface 503 can include a display screen, and optionally can further include a standard wired interface and a wireless interface.

[0138] The network interface 504 can optionally include a standard wired interface and a wireless interface (e.g., a WI-FI interface).

[0139] The application further discloses a computer readable storage medium storing a plurality of instructions, which are adapted to be loaded by a processor to execute the oil-water well control property determination method based on the fuzzy analytic hierarchy process.

[0140] The above are only exemplary embodiments of the present disclosure, and cannot limit the scope of the present disclosure. Any equivalent changes and modifications made in accordance with the teachings of the present disclosure are still within the scope of the present disclosure.

[0141] The present application is intended to cover any variations, uses, or adaptive changes to the general principles of the present disclosure and include common knowledge or conventional technical means in the technical field not recorded in the present disclosure. The scope and spirit of the present disclosure are defined by the claims.

Claims

1. A fuzzy analytic hierarchy process-based oil-water well deliverability property determination method, characterized in that, The method comprises the following steps: S1: obtaining production dynamic index data and preprocessing, constructing a sample data set; S2: according to the actual liquid lifting or water control attribute, the sample data set is divided into two categories, and single factor variance analysis is carried out respectively to obtain a weight coefficient vector; S3: based on the weight coefficient vector, the production dynamic index is respectively constructed into the fuzzy membership function of the oil well end liquid lifting and the oil well end water control two attributes, and the liquid lifting membership degree and the water control membership degree are obtained; Step S3 comprises: S31: a Gaussian function is used to determine the fuzzy membership function; the fuzzy membership function comprises: a liquid lifting membership function and a water control membership function; the sample data set comprises: a liquid lifting well sample and a water control well sample; S32: for each production dynamic index, the fuzzy grade interval is divided according to the value distribution of the liquid lifting well sample and the water control well sample respectively, and the lower quartile q1, the median q2 and the upper quartile q3 of the index in the two kinds of well samples are obtained; The median q2 is selected as the center value c of the Gaussian membership function, and a preset proportion of the quartile interval is selected as the estimation of the standard deviation σ; The expression form of the Gaussian membership function is: wherein is a standardized value of the index, is a center parameter of the index for the target class, is a width parameter; S33: the Gaussian membership function is used to map the continuous numerical production dynamic index into the membership degree value of the liquid lifting attribute and the water control attribute; S4: using the difference comparison fuzzy decision rule, combining the fuzzy membership function, the liquid lifting and water control attributes of the oil and water well are classified.

2. The oil-water well deliverability property determination method based on fuzzy analytic hierarchy process according to claim 1, characterized in that, Step S1 comprises: The daily production dynamic index data is extracted from the oil well and water injection well production database of the target oilfield block and preprocessed to construct a sample data set; the production dynamic index data comprises: daily oil production, daily liquid production, daily water production, water cut, single layer oil production and oil production intensity; The preprocessing step comprises: performing abnormal correction and scale transformation on the production dynamic index data; The abnormal correction of the production dynamic index data comprises: using the quantile truncation method to process the abnormal values of the production dynamic index data, setting the predetermined lower quantile point Ql and the upper quantile point Qu, and adjusting the abnormal values exceeding the [Ql, Qu] interval to the corresponding boundary value; The scale transformation of the production dynamic index data comprises: The daily oil production, daily liquid production and oil production intensity are first logarithmically transformed and then square rooted; The daily water production and single layer oil production are logarithmically transformed; The water cut is processed by inverse complement and logarithmic transformation.

3. The oil-water well deliverability property determination method based on fuzzy AHP of claim 1, wherein, Step S2 comprises: S21: Calculate the single factor variance analysis effect quantity of each production dynamic index and convert to get dimensionless f value; f value is used to represent the discrimination strength of production dynamic index; S22: all f values are normalized to generate an initial weight vector; S23: according to the initial weight vector, combining the index discriminant ratio, a judgment matrix of the analytic hierarchy process is constructed, the relative importance of any two indexes in the judgment matrix is determined by the normalized discriminant ratio, and is mapped to a 1-9 scale; The maximum eigenvalue and consistency ratio CR of the judgment matrix are calculated, when CR<0.1, the consistency of the judgment matrix is acceptable; if CR exceeds 0.1, according to the deviation, the index weight ratio is adjusted or the highly correlated indexes are removed, and then the calculation is carried out until the consistency requirement is met; S24: The weight coefficient vector W of each production dynamic index is calculated by using the geometric mean method combined with the judgment matrix, and the weight coefficient vector is used to represent the relative influence of each index in the lifting / water control decision.

4. The oil-water well deliverability property determination method based on fuzzy AHP of claim 1, wherein, Step S33 includes: For each well to be determined, the production dynamic indexes of the well are substituted into the corresponding lifting membership function and water control membership function to obtain the fuzzy membership matrix of the well relative to the lifting attribute and water control attribute. According to the weight coefficient vector, the membership matrix is weighted and summed to calculate the membership of the well to the liquid lifting attribute and the membership to the water control attribute .

5. The oil-water well deliverability property determination method based on fuzzy AHP of claim 1, wherein, Step S4 includes: Subordinate degree of each well liquid production Subordinate degree of water control Carrying out comparative analysis, according to pre-set fuzzy judgment rule, finally judging the lifting control property of oil-water well, obtaining judgment result, specifically including: calculating the membership difference value and judging according to a preset experience threshold value: When greater than a positive threshold the well is determined to be a liquid lifting well; When Less than a negative threshold The well is determined as a water control well; When the absolute value of the difference between the two values is lower than a preset threshold , then the well is determined to be stable.

6. The oil-water well deliverability property determination method based on fuzzy AHP of claim 1, wherein, The method further includes a batch processing and model adaptive adjustment mechanism and a periodic correction. The batch processing and model adaptive adjustment mechanism specifically includes batch determination and analysis of multiple oil-water wells, automatic extraction of the latest effective production data snapshot of each well for the corresponding period by the pre-set well list and target date batch, and sequential data preprocessing, weight calculation, membership evaluation and lifting / water control determination according to steps S1-S4, batch output of the lifting / water control membership results and recommended control categories of each well. The periodic correction includes periodic correction of the entire algorithm model using real-time updated production dynamic data, feedback adjustment of the weight coefficient vector W and the pre-set empirical threshold δ according to the well condition changes.

7. An electronic device, comprising: The electronic device includes a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to enable the electronic device to perform the oil-water well lifting / water control property determination method based on the fuzzy AHP as claimed in any one of claims 1-6.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores instructions, when the instructions are executed by a computer, a method for determining the lifting / water control property of an oil-water well based on the fuzzy AHP as claimed in any one of claims 1-6 is executed.

Images