Separated layer water injection regulation and control system and method for water injection well

Abstract

The invention relates to the field of oilfield water injection development, and provides a layered water injection regulation and control system and method for a water injection well, and the method comprises the steps: firstly obtaining different types of historical production parameter data of the water injection well, and constructing a stratum parameter confidence factor representing stratum real response characteristics; correcting the statistical characteristics of the historical data by using the confidence factor so as to construct a production parameter weight comparison relationship; then real-time production parameter data is acquired, and the control improvement property of each layer section is identified in combination with a weight comparison relationship; and finally, the target water distribution amount is calculated based on the recognition result, and layered regulation and control are implemented by adjusting the opening degree or injection pressure of a water distributor. According to the method, the problem that the improvement and control property judgment is inaccurate due to data interference in a traditional single-factor variance analysis method is solved, and the accuracy and effectiveness of layered injection allocation of the water injection well are improved.

Description

Technical Field

[0001] This invention relates to the field of oilfield water injection development technology, specifically to a stratified water injection control system and method for water injection wells. Background Technology

[0002] Most of my country's oilfields are heterogeneous and multi-layered, with varying permeability between different layers and even within a single plane. Water injection is an effective method to improve oilfield production. However, indiscriminate water injection methods suffer from low water utilization and poor displacement effects due to varying permeability between layers. Layered water injection, using packers to separate oil layers with significant permeability differences and distribute water accordingly, can effectively regulate water injection conflicts within layers and increase the crude oil recovery rate.

[0003] As oilfields enter the ultra-high water-cut stage, the number and finer details of injection wells in stratified water injection increase, leading to greater inter-layer conflicts. Traditional stratified water injection control methods struggle to accurately consider the impact of layer property combinations and water distribution on the overall well water injection effect, resulting in unreasonable water distribution and affecting reservoir development efficiency. Therefore, a scientific and systematic method for judging the properties of different layers is urgently needed to improve the accuracy and effectiveness of stratified water injection allocation in injection wells. Summary of the Invention

[0004] This invention provides a stratified water injection control system and method for water injection wells. Traditional one-way ANOVA methods suffer from inaccurate judgments due to data interference in determining the control properties of different water injection well strata, leading to unreasonable water injection volume allocation. The specific technical solution adopted is as follows:

[0005] One embodiment of the present invention discloses a stratified water injection control method for a water injection well, the method comprising the following steps:

[0006] In a first aspect, one embodiment of the present invention provides a method for stratified water injection control in a water injection well, the method comprising the following steps:

[0007] Obtain historical production parameter data for different types of water injection wells;

[0008] Construct a formation parameter confidence factor for each data point in the historical production parameter data, excluding the control type data. The formation parameter confidence factor is used to characterize the reliability of the corresponding data point in reflecting the true response characteristics of the formation.

[0009] The statistical characteristics of the historical production parameter data are corrected using the formation parameter confidence factor to obtain the corrected statistical characteristics of the historical production parameter data, and a production parameter weight comparison relationship is constructed based on the corrected statistical characteristics.

[0010] Acquire real-time production parameter data of different types of injection wells to be regulated, compare the real-time production parameter data with historical production parameter data, and identify the lifting and control characteristics of each layer of the injection well to be regulated by combining the weight comparison relationship of the production parameters.

[0011] Based on the identified lifting and control characteristics, the target water distribution volume for each layer of the injection well to be regulated is calculated, and the injection well is stratified and regulated according to the target water distribution volume.

[0012] Preferably, the process of acquiring the historical production parameter data includes:

[0013] The production parameters of different types of water injection wells are collected at a set frequency to obtain the original production parameter data of different types of water injection wells in the past set time period. The original production parameter data includes at least: water injection well formation control properties, daily fluid production, daily oil production, daily water production, water cut, formation pressure, and fluid production intensity.

[0014] The original production parameter data, excluding the water injection well layer lifting and control properties, are normalized to obtain historical production parameter data for different types of water injection wells.

[0015] Preferably, the construction of the stratigraphic parameter confidence factor for each data point in the historical production parameter data, excluding the control type data, includes:

[0016] Based on the control property data, historical production parameter data of the same control property category in the same segment are obtained as an analysis production parameter data set. Based on the difference between each data point in the historical production parameter data and other data points in the analysis production parameter data set, the intra-class deviation of each data point in the historical production parameter data is determined.

[0017] Determine the relevant historical production parameter data of the arbitrary historical production parameter data, and based on the correlation between the local data segment where each data point in the historical production parameter data is located and the corresponding local data segment in the relevant historical production parameter data, determine the inter-class coupling deviation corresponding to each data point in the historical production parameter data;

[0018] The stratigraphic parameter confidence factor for each data point in the historical production parameter data is calculated based on the intra-class bias and inter-class coupling bias. The stratigraphic parameter confidence factor is negatively correlated with the intra-class bias and negatively correlated with the inter-class coupling bias.

[0019] Preferably, calculating the intra-class bias for each data point includes:

[0020] For any data point in the historical production parameter data, obtain the time-continuous data sequence of the same category in the analysis production parameter data set;

[0021] Determine the length of the consecutive data sequences of the same category:

[0022] If the length is less than a preset threshold, the absolute value of the difference between the data point and the mean of all data in the same category data sequence is calculated, and the absolute value of the difference is used as the intra-class deviation of the data point.

[0023] If the length is greater than or equal to the preset threshold, then multiple analysis time windows of different lengths are defined on the continuous data sequence of the same category, and the subsequence corresponding to each analysis time window is obtained;

[0024] For each subsequence corresponding to each analysis time window, the absolute value of the difference between the data point and the mean of all other data in the subsequence is calculated to obtain the corresponding time series deviation.

[0025] Calculate the mean of the temporal deviations corresponding to all time windows, and use the mean as the intra-class deviation for that data point.

[0026] Preferably, the relevant historical production parameter data for determining the arbitrary historical production parameter data includes:

[0027] Calculate the Pearson correlation coefficient between different production parameter data sequences;

[0028] Any two production parameter data whose absolute value of the Pearson correlation coefficient is greater than a preset threshold are recorded as relevant historical production parameter data.

[0029] Preferably, the calculation of the inter-class coupling deviation between the data and the relevant production parameter data includes:

[0030] For any data point in the historical production parameter data, a time window of a preset length is constructed containing the data point, and the difference between each data point in the historical production parameter data and the data point is calculated within the time window to obtain the time neighborhood difference sequence of the data point.

[0031] Calculate the degree of difference between the time neighborhood difference sequence of the arbitrary data point and the time neighborhood difference sequence of the data points at the same time in the relevant historical production parameter data;

[0032] Based on the degree of difference corresponding to all relevant historical production parameter data, the inter-class coupling deviation of any data point is determined.

[0033] Preferably, the step of calculating the stratigraphic parameter confidence factor for each data point in the historical production parameter data based on the intra-class deviation and inter-class coupling deviation includes:

[0034] The intra-class deviation and inter-class coupling deviation are weighted and fused to obtain the comprehensive deviation.

[0035] The overall deviation is negatively correlated and normalized to obtain the formation parameter confidence factor for each data point.

[0036] Preferably, the step of correcting the statistical characteristics of the historical production parameter data using the formation parameter confidence factor to obtain corrected statistical characteristics of different types of historical production parameter data includes:

[0037] Using the formation parameter confidence factor corresponding to each data point as a weight, the weighted mean of the historical production parameter data is calculated to obtain the mean correction result.

[0038] Based on the mean correction results, the statistical characteristics in the one-way ANOVA are calculated to obtain the corrected statistical characteristics.

[0039] Preferably, the step of constructing the production parameter weight comparison relationship based on the statistical features includes:

[0040] Based on one-way ANOVA, the F-values ​​and contribution rates of each production parameter are extracted from the modified statistical features.

[0041] Based on the F-value and contribution of each production parameter, calculate the comprehensive score for each production parameter;

[0042] Based on the differences in comprehensive scores among different production parameters, a control property judgment matrix is ​​constructed; wherein, the element values ​​in the control property judgment matrix are determined by the ratio of the comprehensive scores of the corresponding production parameters.

[0043] Preferably, the identification of the lifting and control properties of each section of the injection well to be regulated includes:

[0044] Based on the weight comparison relationship of the production parameters, the weight coefficients corresponding to each real-time production parameter are determined.

[0045] Based on the weighting coefficients, the real-time production parameter data is compared with the historical production parameter data of each lifting control nature category to obtain the membership degree of each layer to different lifting control nature categories;

[0046] Based on the principle of maximum membership, the lifting and control nature category of each layer is determined.

[0047] Preferably, the step of calculating the target water distribution volume for each section of the injection well to be regulated based on the identified lifting and control characteristics includes:

[0048] Based on Darcy's law and the principle of mass balance, a mathematical model reflecting the relationship between pressure and water injection volume in the strata is established, and a pressure-water injection volume relationship chart is generated.

[0049] Based on the pressure-injection volume relationship chart, and according to the identification results of the lifting and control properties of each layer and the current formation pressure, the target water distribution volume of each layer is determined.

[0050] Preferably, the step of regulating the injection well in layers according to the target water distribution volume includes:

[0051] Adjust the water distributor opening or injection pressure to adjust the actual water injection volume of each layer to the target water distribution volume.

[0052] Secondly, embodiments of the present invention also provide a stratified water injection control system for a water injection well, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0053] This invention offers the following advantages: First, by acquiring historical production parameter data for different types of injection wells and analyzing this data, a formation parameter confidence factor characterizing the true response of the formation is constructed. This confidence factor is then used to correct the statistical characteristics of the historical data, thereby establishing a weighted comparison relationship for production parameters. This reduces the interference from biased data points and weakens the role of these biased data in determining the lifting and control properties. Next, real-time production parameter data for different types of injection wells to be regulated are acquired. This real-time production parameter data is compared with historical production parameter data, and the lifting and control properties of each segment of the injection well are identified based on the weighted comparison relationship. Finally, the target water distribution is calculated based on the identification results, and stratified control is implemented by adjusting the distributor opening or injection pressure. This invention solves the problem of inaccurate judgment of lifting and control properties caused by data interference in traditional one-way ANOVA, improving the accuracy and effectiveness of stratified injection allocation in injection wells. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is a schematic flowchart illustrating the steps of a stratified water injection control method for a water injection well, as provided in an embodiment of the present invention.

[0056] Figure 2 This is a pressure-water injection volume relationship diagram provided for one embodiment of the present invention. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Please see Figure 1 The diagram illustrates a flow chart of a stratified water injection control method for a water injection well according to an embodiment of the present invention. The method includes the following steps:

[0059] Step S100: Obtain historical production parameter data for different types of water injection wells.

[0060] During water injection development in oilfields, due to the heterogeneity of the formation and the differences in geological characteristics of each section, the responses of different sections of the injection well to water injection vary. Single-type production parameter data cannot comprehensively and accurately characterize the true water injection status and formation response characteristics of a section. For example, daily fluid production mainly reflects the production capacity of the corresponding section, while water cut reflects the water drive effect and oil-water distribution of the corresponding section.

[0061] Therefore, in order to accurately identify and further regulate the lifting and control properties of each layer of the injection well, this invention needs to collect various types of historical production parameter data.

[0062] In one possible implementation, acquiring historical production parameter data for different types of injection wells includes: collecting production parameters for different types of injection wells at a set frequency to obtain raw production parameter data for the injection wells over a set time period in the past; the raw production parameter data includes at least: injection well formation control properties, daily fluid production, daily oil production, daily water production, water cut, formation pressure, and fluid production intensity; and normalizing the raw production parameter data to obtain historical production parameter data for different types of injection wells.

[0063] In a specific example, the data collection frequency is set to record production parameter data every 15 minutes, and the time period is set to include water injection well production data from the past 30 days to ensure the timeliness and continuity of the data for subsequent statistical analysis. Implementers can adjust the collection frequency and time period length appropriately based on the specific operating conditions of the oilfield's water injection wells and the data analysis needs.

[0064] The original production parameter data includes at least the following seven types of key parameters:

[0065] Injection Well Interval Control Characteristics: Indicates the control demand category of each interval of the injection well under the current injection state, including three characteristics: water lifting, water control, and water stabilization; Daily Liquid Production of Corresponding Production Well: The total amount of liquid produced daily by each interval of the injection well; Daily Oil Production of Corresponding Production Well: The amount of crude oil produced by production wells with injection-production relationships corresponding to this injection well interval; Daily Water Production of Corresponding Production Well: The amount of water produced daily by each interval of the injection well; Water Cut of Corresponding Production Well: The percentage of water in the produced fluid; Formation Pressure: The formation pressure value corresponding to each interval of the injection well; Production Intensity of Corresponding Production Well: The amount of liquid produced per unit effective thickness of oil layer by production wells with injection-production relationships corresponding to this injection well interval.

[0066] It should be noted that the above seven types of production parameters are preferred embodiments of the present invention. In practical applications, implementers may add or reduce relevant parameter types according to the geological characteristics and production process of specific injection wells. As long as they can reflect the water injection effect and formation response characteristics of each layer of the injection well, these variations fall within the protection scope of the present invention.

[0067] The original production parameter data is normalized to obtain historical production parameter data for different types of injection wells. Because the dimensions and numerical ranges of different types of production parameter data differ significantly (e.g., daily fluid production is in m³ / d, formation pressure is in MPa, and water cut is a percentage), direct statistical analysis would lead to parameters with larger dimensions dominating the analysis results. To eliminate the influence of dimensions and ensure all parameters are on the same order of magnitude for easier subsequent calculations, the original production parameter data needs to be normalized.

[0068] Specifically, the raw production parameter data is divided into two categories:

[0069] Numerical parameters include the daily fluid production, daily oil production, daily water production, water cut, formation pressure, and production intensity of the corresponding production well. These parameters are continuous values ​​and need to be normalized.

[0070] Category label: Water injection well interval control properties (water lifting, water control, water stabilization). This parameter is a classification label used for subsequent group analysis and is not included in normalization calculations.

[0071] For each layer of each injection well, the various numerical production parameters are normally calculated independently. For each type of numerical parameter, the maximum and minimum value normalization method is used for processing.

[0072] By combining the normalized numerical parameter data with the unnormalized categorical labels (injection well layer control properties), we can obtain historical production parameter data of different types of injection wells that can be used for subsequent analysis.

[0073] It should be noted that the above-mentioned maximum and minimum value normalization method is a preferred embodiment of the present invention. It is simple to calculate and can effectively preserve the distribution relationship of the original data. Those skilled in the art can also use other standardization methods (such as Z-score standardization, decimal scaling standardization, etc.) for data processing. As long as the purpose of eliminating dimensions and making the data within a comparable range can be achieved, these variations all fall within the protection scope of the present invention.

[0074] This allows us to obtain historical production parameter data for different types of injection wells, laying a data foundation for further calculations of the properties of each layer of the injection well.

[0075] Step S200: Construct a formation parameter confidence factor for each data point in the historical production parameter data, excluding the control type data. The formation parameter confidence factor is used to characterize the reliability of the corresponding data point in reflecting the true response characteristics of the formation.

[0076] In actual oilfield water injection development, due to the heterogeneity of the formation, fingering during the injection process, wellbore interference, and human interventions such as unblocking, acidizing, and profile control, the historical production parameter data often contains a certain proportion of biased and interfering data. This biased and interfering data does not accurately reflect the formation response characteristics but is caused by abnormal operating conditions or measurement errors. If directly used for statistical analysis, it will severely affect the accuracy of judging the lifting and control characteristics.

[0077] Traditional one-way ANOVA uses equal weighting when calculating the data mean, which fails to distinguish between normal data and biased data, leading to distorted statistical characteristics. To address this issue, this embodiment proposes constructing formation parameter confidence factors to characterize the reliability of each data point in reflecting the true response characteristics of the formation.

[0078] Constructing the stratigraphic parameter confidence factor for each data point in the historical production parameter data, excluding the control type data, including:

[0079] Step S201: Based on the control property data, obtain historical production parameter data of the same control property category in the same layer as an analysis production parameter data set. Based on the difference between each data point in the historical production parameter data and other data points in the analysis production parameter data set, determine the intra-class deviation of each data point in the historical production parameter data.

[0080] In one possible implementation, for any data point in the historical production parameter data, a temporally continuous data sequence of the same category is obtained from the analysis production parameter data set in which it is located.

[0081] In the process of water injection regulation in oilfields, the control characteristics (including three types: water lifting, water control, and water stabilization) of each injection well section directly reflect the regulation needs of that section under the current water injection state. For the first... For each water injection segment, the control nature category of the segment at different time points is first determined based on historical water injection control records. Then, historical production parameter data belonging to the same control nature category, the same water injection segment, and the same production parameter type are extracted to form an analytical production parameter data set.

[0082] It should be noted that, to ensure the effectiveness of the analysis, the number of data points in each production parameter dataset must be greater than or equal to a set threshold, for example, a value of 4. If, within the same stratum and production parameter type, the number of historical data points for a certain lifting control property category (water lifting, water control, or water stabilization) is lower than the set threshold, then all data points in that category are merged with the data from the other lifting control property category with the largest number of data points within the same stratum and production parameter type, forming a new analytical dataset for subsequent calculation of intra-category deviation.

[0083] Ideally, production parameter data of the same controllable property category should exhibit a relatively stable statistical distribution. However, in actual production, due to abrupt changes in fluid flow states in the formation, such as the formation of dominant water channels, breakthrough of the oil-water front, equipment failure, or measurement errors, some data points may deviate from the normal distribution range, forming biased and interfering data. This embodiment determines the intra-class bias of each data point in the historical production parameter data set based on the difference between each data point and other data points in the same analytical production parameter data set.

[0084] For any data point in the analysis production parameter dataset, calculate its intra-class deviation according to the following steps:

[0085] First, for any data point in the historical production parameter data, obtain the time-continuous data sequence of the same category in the analysis production parameter data set.

[0086] Based on the control property category (water lifting, water control, or water stabilization) of any data point in the historical production parameter data, extract all data points that are continuous in time and belong to the same category from the analysis production parameter data set to form a continuous data sequence of the same category, and obtain the number of data points contained in the sequence, i.e., the sequence length is U.

[0087] Secondly, the intra-class bias of the data point is determined based on the length of the continuous data sequence of the same category;

[0088] If the length is less than a preset threshold, the absolute value of the difference between the data point and the mean of all data in the same category data sequence is calculated, and the absolute value of the difference is used as the intra-class deviation of the data point.

[0089] Specifically, a quantity threshold is set, for example, the value of the set quantity threshold is 4, and the implementer can adjust this value according to the specific application scenario. When the sequence length U is less than the set quantity threshold, the absolute value of the difference between any data point in the historical production parameter data and the arithmetic mean of all data points in the entire continuous data sequence of the same category is calculated, and this absolute value is directly used as the intra-class deviation of the data point.

[0090] If the length is greater than or equal to the preset threshold, then:

[0091] Multiple analysis time windows of different lengths are defined on the continuous data sequence of the same category, and the subsequence corresponding to each analysis time window is obtained; for each subsequence corresponding to the analysis time window, the absolute value of the difference between the data point and the mean of all other data in the subsequence is calculated to obtain the corresponding time series deviation; the mean of the time series deviations corresponding to all time windows is calculated, and the mean is used as the intra-class deviation of the data point.

[0092] Specifically, multiple analysis time windows of different lengths are defined on the continuous data sequence of the same category. Each analysis time window starts from the beginning of the sequence and must include the data points to be evaluated. In a specific embodiment of the invention, the length of the analysis time window is configured as follows: let the initial analysis time window length be... ,in As a floor function, the window length increases by 1 sequentially until the sequence length U is the same. Implementers can adjust the definition rules of window length and number according to specific application scenarios.

[0093] For each subsequence corresponding to each analysis time window, calculate the arithmetic mean of the remaining data points excluding the data points mentioned above;

[0094] Calculate the absolute value of the difference between the data point and the arithmetic mean to obtain the corresponding time series deviation.

[0095] Calculate the arithmetic mean of the time series deviations corresponding to each analysis time window, and use this average value as the intra-class deviation of the data points.

[0096] This yielded the intra-class bias for each historical production parameter data point, providing an important evaluation dimension for subsequent construction of formation parameter factors. This calculation method fully considers the temporal and categorical characteristics of oilfield water injection production data, laying a solid foundation for accurately assessing data reliability.

[0097] Step S202: Determine the relevant historical production parameter data of the arbitrary historical production parameter data; based on the correlation between the local data segment where each data point in the historical production parameter data is located and the corresponding local data segment in the relevant historical production parameter data, determine the inter-class coupling deviation corresponding to each data point in the historical production parameter data.

[0098] In one possible implementation, determining the relevant historical production parameter data of the arbitrary historical production parameter data includes: calculating the Pearson correlation coefficient between different production parameter data sequences; and recording any two production parameter data whose absolute value of the Pearson correlation coefficient is greater than a preset threshold as relevant historical production parameter data.

[0099] Specifically, the Pearson correlation coefficient between different production parameter data sequences is calculated; any two production parameter data with an absolute value of Pearson correlation coefficient greater than a preset threshold are recorded as relevant historical production parameter data; based on the degree of difference corresponding to all relevant historical production parameter data, the inter-class coupling bias of any data point is determined.

[0100] In oilfield water injection production systems, different types of production parameters exhibit inherent physical correlations and statistical dependencies. For example, formation pressure is positively correlated with water injection volume, water cut is negatively correlated with production volume, and wellhead temperature is correlated with fluid velocity. These physical correlations reflect the inherent laws governing formation response during water injection development. Furthermore, under normal production conditions, data sequences of physically correlated production parameters will show statistical correlation. When a data point is disturbed, not only will its own value become abnormal, but its coordinated relationship with related parameters will also be disrupted.

[0101] For any two production parameters, calculate the Pearson correlation coefficient between their standardized (i.e., normalized) sequences. A Pearson correlation coefficient greater than zero indicates a positive correlation between the two production parameters; a Pearson correlation coefficient less than zero indicates a negative correlation; and the larger the absolute value of the Pearson correlation coefficient, the stronger the correlation.

[0102] Any two production parameter data points whose absolute Pearson correlation coefficients are greater than a preset threshold are recorded as relevant historical production parameter data. For any given production parameter, relevant historical production parameter data are defined as all other production parameters whose absolute Pearson correlation coefficients with that parameter are greater than a preset threshold. The optimal threshold value can be determined through extensive field data testing; for example, the preset threshold value can be set to 0.5. Implementers can adjust the threshold within the range of [0.4, 0.7] based on the specific geological characteristics and production conditions of the oilfield.

[0103] In practical applications, common relevant parameter pairs include: water injection rate and formation pressure, which are usually positively correlated (correlation coefficient > 0.6); water cut and oil production, which are usually negatively correlated (correlation coefficient < -0.5); and wellhead temperature and fluid velocity, which are usually positively correlated (correlation coefficient > 0.55).

[0104] In one possible implementation, based on the correlation between the local data segment containing each data point in the historical production parameter data and the corresponding local data segment in the relevant historical production parameter data, the inter-class coupling deviation corresponding to each data point in the historical production parameter data is determined, including:

[0105] First, for any data point in the historical production parameter data, a time window of a preset length containing the data point is constructed, and the difference between each data point in the time window and the data point is calculated to obtain the time neighborhood difference sequence of the data point.

[0106] Specifically, with the first Layers in The first moment Historical production parameter data Centered on a time series, a time window of preset length W is constructed. In this embodiment, after optimization testing, the optimal window length W=13 was determined. Implementers can adjust the window length within the range of [9,17] according to specific application scenarios. It should be noted that if the data points are located near the boundaries of the time series, a complete time window cannot be constructed. In this case, an asymmetric window or a method to reduce the window length is used for processing.

[0107] For each data point within the complete time window, calculate the distance between each data point and the center point. The difference is used to obtain the time neighborhood difference sequence; for each relevant historical production parameter data of the j-th production parameter, its data points at the same time are obtained. And construct a time neighborhood difference sequence in the same way.

[0108] Next, the degree of difference between the time neighborhood difference sequence of the arbitrary data point and the time neighborhood difference sequence of the data points at the same time in the relevant historical production parameter data is calculated.

[0109] Specifically, the time neighborhood difference sequence of any data point and the time neighborhood difference sequence of data points at the same moment in the relevant historical production parameter data are input into the difference calculation function, and the function output value is used as the difference.

[0110] It should be noted that the dissimilarity calculation function is used to quantify the similarity between two sequences; a higher dissimilarity indicates a greater difference in the change patterns of the two sequences. This embodiment provides two dissimilarity calculation methods: Euclidean distance and dynamic time warping. Implementers can choose according to their specific needs. In this embodiment, the Euclidean distance method is preferred due to its simplicity and efficiency. When there is a significant time lag effect between production parameters, the dynamic time warping method can be used for calculation.

[0111] Finally, based on the degree of difference corresponding to all relevant historical production parameter data, the inter-class coupling deviation of any data point is determined.

[0112] Inter-class coupling bias is used to quantify the degree to which the coordinated change relationship between a single data point and its related production parameter data is disrupted. Under normal production conditions, correlated production parameters should exhibit coordinated change characteristics. When a data point is disturbed, even a small disturbance can disrupt this coordinated relationship. The calculation formula is as follows:

[0113]

[0114] in, This represents the time neighborhood difference sequence of the nth historical production parameter data at time t for the kth layer segment; This represents the time neighborhood difference sequence of the v-th related historical production parameter data at time t for the k-th segment; This represents the total number of categories of relevant historical production parameter data. This represents the inter-class coupling deviation of the nth historical production parameter data of the kth layer at time t; D is the difference function.

[0115] At this point, the inter-class coupling bias of each historical production parameter data point has been obtained. Together with the intra-class bias calculated in step S201, these two dimensions constitute the two evaluation dimensions of the formation confidence factor. These two dimensions ensure the comprehensiveness and accuracy of the data analysis, laying the data foundation for the subsequent construction of the lifting and control property judgment matrix.

[0116] Step S203: Calculate the stratigraphic parameter confidence factor for each data point in the historical production parameter data based on the intra-class deviation and inter-class coupling deviation, wherein the stratigraphic parameter confidence factor is negatively correlated with the intra-class deviation and negatively correlated with the inter-class coupling deviation.

[0117] In one possible implementation, the intra-class deviation and inter-class coupling deviation are weighted and fused to obtain the comprehensive deviation.

[0118] During water injection development, historical production parameter data are affected by various interference factors, including instrument errors, communication anomalies, and abrupt formation changes, resulting in inconsistent data quality. To accurately assess the lifting and control properties of the formation, it is necessary to quantify the reliability of each data point. Specifically, for each data point of the historical production parameter data, the intra-class deviation and inter-class coupling deviation are calculated according to steps S201 and S202, respectively, and the calculation results are mapped to the [0,1] interval using the maximum and minimum value normalization method. Since the intra-class deviation reflects the statistical deviation of the data point from the same type of data, and the inter-class coupling deviation reflects the disruption of the physical correlation between the data point and related parameters, a comprehensive deviation is obtained through weighted fusion.

[0119] Specifically, for the j-th historical production parameter data point of the i-th layer segment at time t, its intra-class deviation is denoted as... Inter-class coupling deviation is denoted as The overall deviation of the data point is calculated using a weighted linear fusion method. :

[0120]

[0121] in, , For the preset weighting coefficients, satisfy Its size can be set by the implementer according to the specific implementation scenario; in this embodiment, it is taken as... This indicates that both are given equal importance; This represents the intraclass deviation of the j-th historical production parameter of the i-th layer at time t; This represents the inter-class coupling deviation of the j-th historical production parameter of the i-th layer segment at time t.

[0122] Intra-class deviation quantifies the degree of deviation of a data point from other data points within its corresponding control category. Under normal production conditions, data points within the same control category should exhibit similar statistical characteristics. A larger intra-class deviation indicates potential measurement errors, instrument malfunctions, or transient interference, resulting in lower data reliability. Conversely, inter-class coupling deviation quantifies the degree of disruption of the coordinated change relationship between a data point and its related production parameters. Under normal production conditions, physically correlated production parameters should exhibit coordinated change characteristics. A larger inter-class coupling deviation indicates potential local formation anomalies or fluid property changes that disrupt the physical correlation between data point parameters, also resulting in lower data reliability. Therefore, the overall deviation should be negatively correlated with both intra-class and inter-class coupling deviations. Larger intra-class and inter-class coupling deviations indicate greater interference from historical production parameter data, which in turn has a greater impact on the control properties of the assessed stratigraphic segment. Consequently, the weights in the weighted fusion calculation should be smaller.

[0123] Therefore, the formation parameter confidence factor and the overall deviation The correlation is negative, and the comprehensive deviation is transformed into a negative correlation to obtain the formation parameter confidence factor for each data point.

[0124] One possible implementation is as follows:

[0125] First, the maximum and minimum value normalization calculation method is used to process the comprehensive deviation of all data points of the same layer and the same type of production parameter, and map them to the [0,1] interval.

[0126] Then, the normalized comprehensive deviation is subjected to negative correlation transformation, and the mathematical formula for the negative correlation transformation is: , This is the result of the negative correlation transformation. This transformation process achieves a negative correlation, meaning the greater the overall bias, the smaller the initial confidence value. The result after the negative correlation transformation. This is denoted as the formation parameter confidence factor.

[0127] The formation confidence factor is used to quantify the reliability of each historical production parameter data point. This factor comprehensively considers the intra-class consistency and inter-class synergy of the data points. The formation parameter confidence factor reflects the probability of the actual formation response and can be further used to construct accurate formation parameter identification models, optimize water injection control strategies, and predict formation dynamic change trends.

[0128] Thus, a complete calculation process from intra-class deviation and inter-class coupling deviation to formation parameter confidence factors has been realized. The formation parameter confidence factors for each data point are obtained through a weighted fusion comprehensive evaluation method, which significantly improves the accuracy and reliability of the assessment of the lifting and control properties of the layer and provides a high-quality data foundation for water injection regulation optimization.

[0129] Step S300: The statistical characteristics of the historical production parameter data are corrected using the formation parameter confidence factor to obtain the corrected statistical characteristics of the historical production parameter data, and a production parameter weight comparison relationship is constructed based on the corrected statistical characteristics.

[0130] In one possible implementation, obtaining corrected statistical features of different types of historical production parameter data includes using the stratigraphic parameter confidence factor corresponding to each data point as a weight to calculate the weighted mean of the historical production parameter data, and obtaining a mean correction result; based on the mean correction result, recalculating the statistical features in the one-way ANOVA to obtain corrected statistical features.

[0131] Specifically, for any type of production parameter, such as daily liquid production, all its historical data are grouped according to their respective lifting and control categories. For each data point within a group, the corresponding formation parameter confidence factor is directly used as its initial weight. Simultaneously, to ensure a reasonable weighted calculation scale, the weights need to be normalized (e.g., using a sum normalization method) before calculating the mean for each category and the overall dataset, so that within any calculation range, such as within a category or the entire dataset, the sum of the weights used is 1.

[0132] Calculate the sum of the products of all data point values ​​and their normalized weights within each control property category, and record it as the weighted mean of each category; calculate the sum of the products of all data point values ​​and their normalized weights within the overall dataset, and record it as the overall weighted mean.

[0133] Based on the weighted mean of each category and the overall weighted mean, the mean correction result is calculated. Based on the mean correction result, the statistical characteristics in the one-way ANOVA are recalculated, resulting in the following three types of corrected statistical characteristics:

[0134] The modified within-group sum of squares measures the dispersion of data within the same controllability category. For each data point, the difference between its value and the weighted mean of the category is calculated. This difference is squared and multiplied by the normalized weight of the data point to obtain its weighted contribution. The weighted contributions of all data points are summed to obtain the modified within-group sum of squares. The larger this value, the greater the fluctuation of the data within the category around its center.

[0135] The modified between-group sum of squares measures the degree of difference between different control property categories. For each control property category, the difference between the weighted mean of that category and the overall weighted mean is calculated. This difference is squared and multiplied by the total weight of that category, which is the sum of the normalized weights of all data points within that category, to obtain the weighted contribution of that category. The weighted contributions of all categories are summed to obtain the modified between-group sum of squares. The larger this value, the more significant the difference in central values ​​between different categories.

[0136] The adjusted total sum of squares measures the overall dispersion of data around the population center. For each data point, the difference between its value and the population weighted mean is calculated. This difference is squared and multiplied by the normalized weight of that data point to obtain its weighted contribution. The weighted contributions of all data points are summed to obtain the adjusted total sum of squares. According to statistical principles, the adjusted total sum of squares equals the sum of the adjusted between-group sum of squares and the adjusted within-group sum of squares.

[0137] Finally, the corrected statistical features are output. The calculated corrected within-group sum of squares, corrected between-group sum of squares, and corrected total sum of squares are used as the corrected statistical features of the historical data for this production parameter type. These statistical features will then be directly used to calculate the F-value and contribution of each production parameter, thereby constructing a weighted comparison relationship between production parameters.

[0138] In one possible implementation, the production parameter weight comparison relationship is constructed based on the modified statistical features, including extracting the F-value and contribution of each production parameter from the modified statistical features based on one-way ANOVA; calculating the comprehensive score of each production parameter based on the F-value and contribution of each production parameter; and constructing a control property judgment matrix based on the differences in comprehensive scores between different production parameters.

[0139] Specifically, as shown in Table 1, for each historical production parameter data, the F-value was calculated based on one-way ANOVA using the adjusted between-group mean square and within-group mean square. This value reflects the statistical significance of the production parameter in distinguishing different categories of water control properties (water lifting, water control, and water stabilization). The larger the F-value, the more significant the difference between different categories of water control properties, and the stronger the ability to distinguish between categories. The contribution was calculated based on the adjusted between-group mean square and the sum of squared deviations. This value reflects the proportion of the production parameter that explains the overall variation in production data. The greater the contribution, the more information the parameter contains in the judgment of controllability, and the more important its contribution to the judgment result.

[0140] Table 1 Results of One-Way ANOVA

[0141]

[0142] It should be noted that the F-value is related to the degree of contribution. The calculation of is a well-known technique in one-way ANOVA, and its specific calculation process is well known to those skilled in the art, so it will not be elaborated here.

[0143] Based on the F-values ​​and contributions of each production parameter, a comprehensive score is calculated for each parameter. This comprehensive score quantifies the overall importance of each production parameter in the judgment of control properties. In one embodiment, the F-value and contribution are directly multiplied to obtain the comprehensive score. In another embodiment, weights for the F-value and contribution are pre-set, such as both weights being 0.5. These weights are then used to normalize the F-value and contribution (e.g., maximum / minimum value normalization), and the weighted sum is used as the comprehensive score. This step integrates the F-value, reflecting the parameter's discriminative ability, and the contribution, reflecting the parameter's explanatory ability, into a unified score.

[0144] Based on the comprehensive score differences among different production parameters, a control property judgment matrix is ​​constructed. This matrix is ​​the mathematical expression of the weight comparison relationship of the production parameters. Let the total number of production parameter types be M, then the constructed control property judgment matrix A is an M×M matrix. The element in the p-th row and q-th column of the matrix... This characterizes the weight comparison relationship between the p-th production parameter and the q-th production parameter in the judgment of control properties. Its value is determined by the ratio of their comprehensive importance scores, i.e. ;in, and These are the comprehensive scores for the p-th and q-th production parameters, respectively. The control property judgment matrix comprehensively depicts the relative importance ranking ratio between all production parameters, providing a core weighting basis for subsequent identification of layer control properties based on fuzzy comprehensive evaluation.

[0145] Thus, the control property judgment matrix was constructed, completing the calculation process from correcting statistical characteristics to comparing the weights of production parameters. By introducing formation parameter confidence factors to correct the traditional one-way ANOVA, and on this basis, the statistical characteristics of each production parameter were extracted and integrated, ultimately forming a weight comparison relationship that accurately reflects the relative importance of each production parameter in the control property judgment. This lays a crucial foundation for the subsequent accurate identification of the control properties of injection well intervals.

[0146] Step S400: Obtain real-time production parameter data of different types of injection wells to be regulated, compare the real-time production parameter data with historical production parameter data, and identify the lifting and control characteristics of each layer of the injection well to be regulated by combining the production parameter weight comparison relationship.

[0147] In one possible implementation, the weight coefficients corresponding to each real-time production parameter are determined based on the weight comparison relationship of the production parameters; based on the weight coefficients, the real-time production parameter data and the historical production parameter data of each control property category of the weight coefficients are compared to obtain the membership degree of each layer to different control property categories; and the control property category to which each layer belongs is determined according to the principle of maximum membership degree.

[0148] First, for the target layer of the injection well to be regulated, real-time production parameter data is acquired using the same acquisition frequency and type as historical data. The type of the real-time production parameter data should be consistent with the historical data. To ensure data comparability, these real-time data need to undergo the same normalization process as the historical data. For example, a maximum-minimum normalization method is used, and the average of the normalized production parameter data of the same type is calculated to obtain the real-time production parameter vector, denoted as [vector]. Where M is the total number of production parameter types.

[0149] Next, the control property judgment matrix constructed in step S300 is obtained, and the weight coefficients of each historical production parameter data are extracted from the control property judgment matrix. The control property judgment matrix is... The square array whose elements Characterized the first The production parameters relative to the first The importance ratio of various production parameters.

[0150] For use in subsequent calculations, a weight coefficient vector reflecting the absolute importance of each parameter needs to be derived from this relative importance matrix. And satisfy A specific method for deriving the weight coefficient vector is the row geometric mean method: calculate the geometric mean of each row element in the control property judgment matrix, perform sum normalization on the geometric mean, that is, divide it by the sum of the geometric means of all rows, and record the normalization result as the weight coefficient of the production parameter of the corresponding category in that row.

[0151] Then, using historical production parameter data and their corresponding formation parameter confidence factors, a reference pattern vector is constructed for each lifting control property category. Let the set of lifting control property categories be... These correspond to water lifting, water control, and water stabilization, respectively. (For each category...) Its reference mode vector is , where M is the total number of production parameter types. Specifically, for the i-th (i=1,2,…M) production parameter, ,in, The total number of data points in the historical production parameter data of the k-th category. For the l-th data point in the historical production parameter data of the k-th category, is the confidence factor for the formation parameters corresponding to the l-th data point.

[0152] Furthermore, for the segment to be determined, the real-time production parameter vector and each reference mode vector are calculated. The weighted Euclidean distance between them ,in, The weighting coefficients are obtained by normalizing the i-th production parameter. To generate the i-th component of the parameter vector in real time, For category The i-th component of the reference pattern vector. Distance. The smaller the value, the closer it is to real-time data and category. The more similar the typical patterns are.

[0153] Finally, the distance is converted into membership degree. The smaller the distance, the higher the similarity between that layer and a certain category, and the greater the membership degree. The specific conversion formula is as follows:

[0154]

[0155] in, This indicates that the segment belongs to a category. membership degree This is a preset constant, set to 0.001 in this embodiment to prevent the denominator from being zero. Thus, the membership vector of this layer to the three control property categories is obtained. And satisfy .

[0156] Finally, based on the principle of maximum membership, the control nature category to which each layer belongs is determined.

[0157] Compare membership vectors The three components are used to determine the control property category corresponding to the largest membership degree as the final identification result for that layer. If there are multiple categories with the same membership degree and all of them being the maximum value, the determination can be made by combining engineering experience or further historical data backtracking, or it can be directly classified as a hidden water property.

[0158] It should be noted that the above-described fuzzy comprehensive evaluation method is a preferred implementation. Those skilled in the art can also use other pattern recognition or classification algorithms (such as K-nearest neighbors, support vector machines, etc.) combined with the weighted comparison relationship of the production parameters to identify the lifting and control properties. These variations all fall within the protection scope of this invention. The lifting and control property judgment matrix is ​​a static knowledge base calculated based on historical data, which can be updated periodically in practical applications to adapt to the dynamic changes in oilfield development. Simultaneously, the preprocessing of real-time production parameter data must be strictly consistent with the preprocessing method of historical data to ensure that the data are compared and calculated at the same scale.

[0159] This completes the automatic identification of the control characteristics of each layer of the injection well to be regulated, providing a reliable basis for the subsequent accurate calculation of the target water distribution volume.

[0160] Step S500: Based on the identified lifting and control properties, calculate the target water distribution volume for each section of the injection well to be regulated, and perform stratified water injection regulation on the injection well according to the target water distribution volume.

[0161] In one possible implementation, based on the identified lifting and control properties, the target water distribution volume for each segment of the injection well to be regulated is calculated, including: establishing a mathematical model reflecting the relationship between segment pressure and injection volume based on Darcy's law and the principle of mass balance, and generating a pressure-injection volume relationship chart; based on the pressure-injection volume relationship chart, and according to the identification results of the lifting and control properties of each segment and the current formation pressure, querying and determining the target water distribution volume for each segment.

[0162] In oilfield water injection development, the appropriate water distribution volume for each formation is closely related to its current formation pressure and lifting / control properties. To scientifically determine the target water distribution volume for each formation, this embodiment establishes a mathematical model reflecting the relationship between formation pressure and water injection volume based on Darcy's law and the principle of mass balance, and generates a pressure-water injection volume relationship chart.

[0163] Specifically, based on Darcy's law and the principle of mass balance, a mathematical model reflecting the relationship between formation pressure and injection volume is established. According to Darcy's linear seepage law, the seepage flow rate q (i.e., injection volume) of fluid in the porous media of the formation is related to the pressure difference. Formation permeability seepage cross-sectional area Proportional to fluid viscosity and seepage path length They are inversely proportional, and the relationship is as follows:

[0164]

[0165] Where q is the water injection volume of the layer ( ); The effective permeability of the layer (mD); A is the equivalent cross-sectional area of ​​the seepage in the layer (mD). ); The difference between bottom hole flowing pressure and formation pressure ; The viscosity of the injected fluid (mPa) s); L is the equivalent seepage path length (m).

[0166] Using the mathematical model, with formation pressure (or pressure difference) () is the horizontal axis, and the water injection volume is the vertical axis. Using the vertical axis as the ordinate, a cluster of pressure-injection rate curves was plotted at different permeability levels, forming a pressure-injection rate graph, as shown below. Figure 2 As shown in the figure, the chart visually reflects the formation pressure (or pressure difference) required to achieve a certain injection volume under specific formation properties, and vice versa. The chart can be generated using numerical simulation software or obtained based on regression analysis of historical injection data.

[0167] Based on the pressure-injection volume relationship chart, and according to the identification results of the lifting control characteristics of each layer and the current formation pressure, the target water distribution volume for each layer is determined. For each layer of the injection well to be regulated, the following operations are performed:

[0168] The formation pressure value is obtained by acquiring the water control properties (water lifting, water control, or water stabilization) identified in step S400 of this formation, as well as the current formation pressure value obtained by real-time monitoring or calculation through downhole pressure gauges. .

[0169] If the identification result is water lifting, it means that water injection needs to be increased in this section to enhance displacement. The target water distribution volume should be increased by a preset ratio (e.g., 10%~30%) based on the current actual water injection volume, or determined according to the target pressure recovery value required by the development plan.

[0170] If the identification result is water control, it means that the water injection in this section needs to be reduced to suppress water cross-flow. The target water distribution should be reduced by a preset ratio (e.g., 10%~30%) based on the current actual water injection volume.

[0171] If the identification result is stable water, it means that the water injection status of this layer is moderate, and the target water distribution can be maintained near the current actual water injection volume, or finely adjusted according to the steady-state injection-production ratio.

[0172] Based on the current formation pressure of this section Given the control direction and amplitude, in the pressure-injection volume relationship chart, find the injection volume value that meets the control requirements at that pressure level, and determine this value as the target water distribution volume for that layer. For example, for the water-lifting layer, look for the pressure in the diagram. It can achieve a water injection volume that is 20% higher than the current injection volume.

[0173] At this point, the target water distribution volume for each section of the injection well to be regulated can be obtained.

[0174] After obtaining the target water distribution volume for each layer, the actual water injection volume needs to be adjusted through the downhole control device. Layered water injection control of the injection well is performed according to the target water distribution volume, including adjusting the opening of the water distributor or the injection pressure to adjust the actual water injection volume of each layer to the target water distribution volume.

[0175] Specifically, adjust the opening of the water distributors installed at corresponding locations in each section, or adjust the injection pressure of that section, so that the actual water injection flow rate of that section is stabilized at the calculated target water distribution volume. nearby.

[0176] In a specific example, the control process can be achieved using an intelligent tiered water injection process string. Each section of this string is equipped with an adjustable distributor (such as an eccentric distributor or an intelligent distribution valve) and a flow / pressure sensor. The central controller receives the target water injection volume command for each section and adjusts the orifice diameter of the corresponding distributor via hydraulic or motor drive, thereby changing the flow area and achieving precise flow control; or it adjusts the injection pressure by regulating the regulating valve of the injection pipeline for that section, thereby adjusting the water injection volume. During the control process, the actual water injection volume of each section is monitored in real time, and a closed-loop control algorithm (such as PID control) is used to make the actual water injection volume quickly and stably approach the target water injection volume.

[0177] It should be noted that the pressure map method is one of the commonly used methods in the field for determining a reasonable water injection volume. Those skilled in the art can also use other methods such as numerical simulation-based optimization algorithms and empirical formulas to calculate the target water distribution volume, as long as they are based on the identification results of the control properties of the strata. These variations all fall within the protection scope of this invention. The correspondence between the specific adjustment amount of the water distributor opening or injection pressure and the target water distribution volume (i.e., the control characteristic curve) can be obtained through indoor experiments, software simulation, or on-site calibration. This invention does not limit the specific execution equipment (water distributor type, drive method, sensor type) for stratified water injection; any existing water injection process tubing capable of independently adjusting the water injection volume by strata is applicable.

[0178] This completes the process from intelligent identification of lifting and control characteristics to calculation of target water distribution volume, and then to precise control of stratified water injection. By improving the accuracy of lifting and control characteristic identification, the rationality of subsequent water distribution volume calculation is fundamentally guaranteed, ultimately achieving efficient and refined development of injection wells.

[0179] Based on the same inventive concept, this invention also provides a stratified water injection control system for a water injection well. The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The system is characterized in that when the processor executes the computer program, it implements the steps of the stratified water injection control method for a water injection well as described above.

[0180] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for layered water injection control in a water injection well, characterized in that, Includes the following steps: Obtain historical production parameter data for different types of water injection wells; Construct a formation parameter confidence factor for each data point in the historical production parameter data, excluding the control type data. The formation parameter confidence factor is used to characterize the reliability of the corresponding data point in reflecting the true response characteristics of the formation. The statistical characteristics of the historical production parameter data are corrected using the formation parameter confidence factor to obtain the corrected statistical characteristics of the historical production parameter data, and a production parameter weight comparison relationship is constructed based on the corrected statistical characteristics. Acquire real-time production parameter data of different types of injection wells to be regulated, compare the real-time production parameter data with historical production parameter data, and identify the lifting and control characteristics of each layer of the injection well to be regulated by combining the weight comparison relationship of the production parameters. Based on the identified lifting and control characteristics, the target water distribution volume for each layer of the injection well to be regulated is calculated, and the injection well is stratified and regulated according to the target water distribution volume.

2. The method for layered water injection control of an injection well according to claim 1, characterized in that, The process of acquiring the historical production parameter data includes: The production parameters of different types of water injection wells are collected at a set frequency to obtain the original production parameter data of different types of water injection wells in the past set time period. The original production parameter data includes at least: water injection well formation control properties, daily fluid production, daily oil production, daily water production, water cut, formation pressure, and fluid production intensity. The original production parameter data, excluding the water injection well layer lifting and control properties, are normalized to obtain historical production parameter data for different types of water injection wells.

3. The method for layered water injection control of an injection well according to claim 1, characterized in that, The construction of the stratigraphic parameter confidence factor for each data point in the historical production parameter data, excluding the lifting control type data, includes: Based on the control property data, historical production parameter data of the same control property category in the same segment are obtained as an analysis production parameter data set. Based on the difference between each data point in the historical production parameter data and other data points in the analysis production parameter data set, the intra-class deviation of each data point in the historical production parameter data is determined. Determine the relevant historical production parameter data of the arbitrary historical production parameter data, and based on the correlation between the local data segment where each data point in the historical production parameter data is located and the corresponding local data segment in the relevant historical production parameter data, determine the inter-class coupling deviation corresponding to each data point in the historical production parameter data; The stratigraphic parameter confidence factor for each data point in the historical production parameter data is calculated based on the intra-class bias and inter-class coupling bias. The stratigraphic parameter confidence factor is negatively correlated with the intra-class bias and negatively correlated with the inter-class coupling bias.

4. The method for layered water injection control of an injection well according to claim 3, characterized in that, The calculation of the intra-class bias for each data point includes: For any data point in the historical production parameter data, obtain the time-continuous data sequence of the same category in the analysis production parameter data set; Determine the length of the consecutive data sequences of the same category: If the length is less than a preset threshold, the absolute value of the difference between the data point and the mean of all data in the same category data sequence is calculated, and the absolute value of the difference is used as the intra-class deviation of the data point. If the length is greater than or equal to the preset threshold, then multiple analysis time windows of different lengths are defined on the continuous data sequence of the same category, and the subsequence corresponding to each analysis time window is obtained; For each subsequence corresponding to each analysis time window, the absolute value of the difference between the data point and the mean of all other data in the subsequence is calculated to obtain the corresponding time series deviation. Calculate the mean of the time series deviations corresponding to all time windows, and use the mean as the intra-class deviation for that data point.

5. The method for layered water injection control of an injection well according to claim 3, characterized in that, The relevant historical production parameter data for determining the arbitrary historical production parameter data includes: Calculate the Pearson correlation coefficient between different production parameter data sequences; Any two production parameter data whose absolute value of the Pearson correlation coefficient is greater than a preset threshold are recorded as relevant historical production parameter data.

6. The method for layered water injection control of an injection well according to claim 3, characterized in that, The calculation of the inter-class coupling deviation between the data and the relevant production parameter data includes: For any data point in the historical production parameter data, a time window of a preset length containing the data point is constructed, and the difference between each data point in the historical production parameter data and the data point is calculated within the time window to obtain the time neighborhood difference sequence of the data point. Calculate the degree of difference between the time neighborhood difference sequence of the arbitrary data point and the time neighborhood difference sequence of the data points at the same time in the relevant historical production parameter data; Based on the degree of difference corresponding to all relevant historical production parameter data, the inter-class coupling deviation of any data point is determined.

7. The method for layered water injection control of an injection well according to claim 3, characterized in that, The step of calculating the stratigraphic parameter confidence factor for each data point in the historical production parameter data based on the intra-class deviation and inter-class coupling deviation includes: The intra-class deviation and inter-class coupling deviation are weighted and fused to obtain the comprehensive deviation. The overall deviation is negatively correlated and normalized to obtain the formation parameter confidence factor for each data point.

8. The method for layered water injection control of an injection well according to claim 1, characterized in that, The step of correcting the statistical characteristics of historical production parameter data using the formation parameter confidence factor to obtain corrected statistical characteristics of different types of historical production parameter data includes: Using the formation parameter confidence factor corresponding to each data point as a weight, the weighted mean of the historical production parameter data is calculated to obtain the mean correction result. Based on the mean correction results, the statistical characteristics in the one-way ANOVA are calculated to obtain the corrected statistical characteristics.

9. The method for layered water injection control of an injection well according to claim 1, characterized in that, The construction of the production parameter weight comparison relationship based on the statistical features includes: Based on one-way ANOVA, the F-values ​​and contribution rates of each production parameter are extracted from the modified statistical features. Based on the F-value and contribution of each production parameter, calculate the comprehensive score for each production parameter; Based on the differences in comprehensive scores among different production parameters, a control property judgment matrix is ​​constructed; wherein, the element values ​​in the control property judgment matrix are determined by the ratio of the comprehensive scores of the corresponding production parameters.

10. The method for layered water injection control of an injection well according to claim 1, characterized in that, The identification of the lifting and control properties of each section of the injection well to be regulated includes: Based on the weight comparison relationship of the production parameters, the weight coefficients corresponding to each real-time production parameter are determined. Based on the weighting coefficients, the real-time production parameter data is compared with the historical production parameter data of each lifting control nature category to obtain the membership degree of each layer to different lifting control nature categories; Based on the principle of maximum membership, the lifting and control nature category of each layer is determined.

11. The method for layered water injection control of an injection well according to claim 1, characterized in that, The calculation of the target water distribution volume for each section of the injection well to be regulated based on the identified lifting and control characteristics includes: Based on Darcy's law and the principle of mass balance, a mathematical model reflecting the relationship between pressure and water injection volume in the strata is established, and a pressure-water injection volume relationship chart is generated. Based on the pressure-injection volume relationship chart, and according to the identification results of the lifting and control properties of each layer and the current formation pressure, the target water distribution volume of each layer is determined.

12. The method for layered water injection control of an injection well according to claim 1, characterized in that, The step of regulating the injection wells in layers according to the target water distribution volume includes: Adjust the water distributor opening or injection pressure to adjust the actual water injection volume of each layer to the target water distribution volume.

13. A stratified water injection control system for a water injection well, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the layered water injection control method for a water injection well as described in any one of claims 1-12.

Images