A Machine Learning-Based Simulation Method and System for Borehole Water Injection Migration
By using machine learning-based methods to collect and analyze multi-source state data during borehole water injection, a dynamic transport simulation model is constructed, which solves the problem of the difficulty in simulating the dynamic evolution process of hydraulic channels in existing technologies, and realizes accurate simulation and analysis of groundwater transport patterns.
Patent Information
- Application Number
- CN202611376170.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-07
- Publication Date
- 2026-11-17
AI Technical Summary
Existing borehole water migration simulation methods are unable to accurately simulate the continuous evolution of the underground state as the hydraulic channels form, expand, and stably migrate during the water injection process, and are also unable to identify the latent state of groundwater migration.
Using a machine learning-based approach, multi-source state data is collected during the borehole water injection process to form a raw state dataset. This dataset is then preprocessed and analyzed for features to construct a dynamic transport simulation model. This model simulates the water migration path and its impact range, generating simulation results of transport patterns.
It has achieved accurate identification of the hidden state of groundwater migration and simulation of the dynamic evolution process of hydraulic channels, and obtained dynamic analysis of the water migration path, influence range and migration trend during the borehole water injection process.
Smart Images

Figure CN123116840A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of groundwater transport simulation, and in particular to a method and system for simulating the transport patterns of borehole water injection based on machine learning. Background Technology
[0002] During borehole water injection, by introducing water at a certain pressure into the underground medium and collecting parameters such as injection pressure, flow rate, and hydraulic response, the fracture structure of underground rock and the characteristics of water migration can be analyzed. With the development of computing technology, existing methods for analyzing groundwater migration during borehole injection are gradually evolving from traditional empirical analysis to numerical simulation methods. By establishing seepage calculation models, fracture network models, and hydraulic conduction models, the diffusion range, migration path, and pressure propagation process of water during injection can be simulated, providing technical support for the study of groundwater migration patterns.
[0003] Existing borehole water injection migration simulation methods typically rely on fixed geological parameters and pre-defined hydraulic channel structures. These methods fall short in representing the continuous evolution of the subsurface state as the hydraulic channels form, expand, and stably migrate during the injection process. Therefore, it is necessary to combine the characteristics of multi-source state data changes during borehole water injection to achieve latent state analysis of groundwater migration and to conduct dynamic migration simulations based on the life cycle changes of the hydraulic channels. Summary of the Invention
[0004] In view of the aforementioned existing problems, the present invention is proposed.
[0005] Therefore, this invention provides a machine learning-based simulation method for groundwater migration during borehole injection, which solves the problems of difficulty in identifying the hidden state of groundwater migration and the difficulty in accurately simulating the dynamic evolution of hydraulic channels during borehole injection.
[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for simulating the migration patterns of borehole water injection based on machine learning, comprising, Collect multi-source state data of borehole water injection during the borehole water injection process, and identify the data according to the collection time and spatial location of the multi-source state data of borehole water injection to form a raw state dataset of borehole water injection. The original state dataset of borehole water injection is preprocessed to form a borehole water injection state database. Feature analysis is then performed on the borehole water injection state database to obtain the hidden state feature information of groundwater migration. Continuous change analysis is performed based on the characteristics of groundwater transport and migration to obtain the life cycle status information of hydraulic channels; Based on the hidden state characteristics of groundwater migration and the life cycle state information of hydraulic channels, a dynamic migration simulation model of borehole water injection is constructed. The dynamic migration simulation model of borehole water injection was used to simulate the water migration path, influence range and migration trend during the borehole water injection process, and the simulation results of borehole water injection dynamic migration were obtained. Based on the simulation results of borehole water injection dynamic migration, simulation results of borehole water injection migration law are generated.
[0007] As a preferred embodiment of the machine learning-based borehole water injection migration law simulation method of the present invention, the method for collecting multi-source state data of borehole water injection during the borehole water injection process includes: Data on changes in water injection pressure, water injection flow rate, cumulative water injection volume, hydraulic response, and geological structure correlation were collected during the borehole water injection process.
[0008] As a preferred embodiment of the borehole water injection migration law simulation method based on machine learning described in this invention, the step of identifying the acquisition time and spatial location corresponding to the multi-source state data of the borehole water injection to form the original state dataset of the borehole water injection includes: Based on the water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data, and geological structure association data in the borehole water injection multi-source state data, the corresponding data acquisition time is recorded as the acquisition time identifier, and a spatial location identifier is generated based on the borehole depth location, spatial coordinate location, and monitoring area location corresponding to the data acquisition location. Based on the acquisition time identifier and the spatial location identifier, the data on changes in water injection pressure, changes in water injection flow rate, cumulative water injection volume, hydraulic response data, and geological structure correlation data corresponding to the same acquisition time and spatial location are matched to form a dataset of the original state of borehole water injection.
[0009] As a preferred embodiment of the machine learning-based borehole water injection migration law simulation method of the present invention, the preprocessing of the original borehole water injection state dataset to form a borehole water injection state database includes: Based on the water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data, and geological structure association data in the original state dataset of borehole water injection, the different types of data are formatted, abnormal data is corrected, and missing data is supplemented to obtain the processed original state dataset of borehole water injection. Based on the processed raw data set of borehole water injection, the time and spatial location information corresponding to the data on changes in water injection pressure, changes in water injection flow rate, cumulative water injection volume, hydraulic response, and geological structure are organized to form a borehole water injection status database.
[0010] As a preferred embodiment of the machine learning-based borehole water injection migration law simulation method of the present invention, wherein: the feature analysis of the borehole water injection state database to obtain the latent state feature information of groundwater migration includes: Based on the water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data, and geological structure correlation data in the borehole water injection status database, the time nodes when the water injection pressure change data changes are obtained, and the time nodes when the hydraulic response data corresponding to different spatial locations reaches the corresponding change state are obtained. The pressure response delay relationship is established based on the time interval between the time nodes of the water injection pressure change data and the hydraulic response data, and the spatial migration delay relationship is established based on the time interval between the time nodes of the hydraulic response data corresponding to different spatial locations, so as to obtain the borehole water injection response delay characteristic information. Based on the groundwater migration state change characteristics, the pressure response delay relationship and spatial migration delay relationship corresponding to different water injection stages are continuously changed and learned. By analyzing the change trend, change magnitude and stage change characteristics of the pressure response delay relationship and spatial migration delay relationship in the continuous water injection stage, when the change magnitude of the pressure response delay relationship and spatial migration delay relationship in the continuous water injection stage is lower than the preset change range, the groundwater is determined to be in the initial response state. in, When the pressure response delay relationship and spatial migration delay relationship show a continuous decreasing change during the continuous water injection stage, it is determined that the groundwater is in the state of hydraulic channel formation and expansion. When the variation amplitudes of pressure response delay relationship and spatial migration delay relationship gradually decrease during the continuous water injection stage and maintain consistent variation characteristics in the continuous stage, it is determined that the groundwater is in a stable migration state. Based on the changing characteristics of pressure response delay and spatial migration delay at different water injection stages, information on the latent state of groundwater migration is obtained.
[0011] As a preferred embodiment of the machine learning-based borehole water injection migration law simulation method of the present invention, wherein: the step of continuously changing the feature information of groundwater migration latent state to obtain the life cycle state information of the hydraulic channel includes: Based on the groundwater migration and hidden state feature information, a groundwater migration and hidden state inheritance sequence is established according to the continuous water injection stage. The correspondence relationship of the groundwater migration and hidden state feature information between adjacent water injection stages is matched, and the groundwater migration and hidden state feature information that exists in all continuous water injection stages is determined as the inheritance state. The groundwater migration and latent state feature information that first appears during the continuous water injection stage is identified as the new state, and the groundwater migration and latent state feature information that disappears during the continuous water injection stage is identified as the reconstructed state. The inherited feature information of the groundwater migration and latent state is obtained based on the inherited state, the new state, and the reconstructed state. Based on the inherited feature information of groundwater migration hidden state, the proportion of the number of groundwater migration hidden state feature information corresponding to the inherited state in the continuous water injection stage to the total number of groundwater migration hidden state feature information is calculated to obtain the inheritance ratio. The reconstruction ratio is obtained by statistically analyzing the proportion of groundwater migration latent state feature information corresponding to the reconstructed state during the continuous water injection stage to the total number of groundwater migration latent state feature information. The characteristic information of the hydraulic channel evolution stage is determined based on the changes in the inheritance ratio and reconstruction ratio during the continuous water injection stage; Based on the characteristic information of the hydraulic channel evolution stage, the inheritance ratio and reconstruction ratio corresponding to the continuous water injection stage are continuously analyzed. When the inheritance ratio increases in the continuous water injection stage, it is determined that the hydraulic channel is in a continuous development process. When the inheritance ratio and reconstruction ratio alternate during the continuous water injection phase, it is determined that the hydraulic channel is in the adjustment process. When the reconstruction ratio increases during the continuous water injection phase, it is determined that the hydraulic channel is in the renewal process. Based on the characteristic information of the hydraulic channel evolution stage corresponding to different water injection phases, the life cycle status information of the hydraulic channel is obtained.
[0012] As a preferred embodiment of the borehole water injection migration law simulation method based on machine learning described in this invention, the construction of the borehole water injection dynamic migration simulation model based on the latent state feature information of groundwater migration and the life cycle state information of hydraulic channels includes: Based on the groundwater migration and hidden state characteristic information, the state change process between adjacent groundwater migration and hidden state characteristic information is extracted according to the continuous water injection stage. The water injection conditions corresponding to the transformation of each groundwater migration and hidden state characteristic information to other groundwater migration and hidden state characteristic information are statistically analyzed. The repeated state change processes under the same water injection conditions are grouped into the same state driving relationship to obtain the groundwater migration and hidden state driving relationship information. Based on the groundwater transport and migration state driving relationship information and the hydraulic channel life cycle state information, the frequency of occurrence of groundwater transport and migration state driving relationships corresponding to different hydraulic channel life cycle state information is counted. Groundwater transport and migration state driving relationships with a frequency higher than a preset frequency threshold are configured as state evolution rules under the corresponding hydraulic channel life cycle state information. The state evolution rules corresponding to different hydraulic channel life cycle state information are combined to obtain a dynamic state evolution rule set. Based on the dynamic state evolution rule set, the hidden state characteristics of groundwater migration are used as the initial state input, and the dynamic state evolution rule set is used as the basis for updating the hidden state of groundwater migration. The hidden state characteristics of groundwater migration after each water injection stage are recursively updated, and a borehole water injection dynamic migration simulation model is constructed according to the recursive results of the hidden state characteristics of groundwater migration.
[0013] As a preferred embodiment of the borehole water injection migration law simulation method based on machine learning described in this invention, the following steps are taken: The borehole water injection dynamic migration simulation model is used to simulate the water migration path, influence range, and migration trend during the borehole water injection process, and the simulation results of the borehole water injection dynamic migration include: Based on the life cycle status information of the hydraulic channel, corresponding migration rules are established for different migration stages in the dynamic migration simulation model of borehole water injection. Different water body migration modes are matched with different life cycle stages corresponding to the life cycle status information of the hydraulic channel to obtain water body dynamic migration rule information. Based on the dynamic migration rules of water bodies, the water migration process in the borehole water injection dynamic migration simulation model is controlled by stage switching. When the life cycle status information of the hydraulic channel changes in a stage, the corresponding dynamic migration rules of water bodies are switched synchronously to update the water migration direction, migration expansion mode and migration advancement process, thereby obtaining dynamic migration process information. Based on the dynamic migration process information, the borehole water injection dynamic migration simulation model is used to continuously generate the water migration path, influence range and migration trend corresponding to different water injection stages, and obtain the borehole water injection dynamic migration simulation results.
[0014] As a preferred embodiment of the borehole water injection migration law simulation method based on machine learning described in this invention, the step of generating borehole water injection migration law simulation results based on the borehole water injection dynamic migration simulation results includes: Based on the dynamic migration simulation results of borehole water injection, the changes in water migration path, influence range, and migration trend corresponding to the continuous water injection stage are extracted. The repeated changes in the continuous water injection stage are classified to obtain migration evolution pattern information. Based on the information of transport evolution patterns, the connection between different transport evolution patterns is continuously tracked. Transport evolution patterns that can be continuously connected are combined into a complete transport process. Transport evolution patterns that are interrupted or replaced are divided into different transport processes to obtain transport pattern association information. Based on the information related to the migration patterns, the water migration paths, influence ranges, and migration trends corresponding to different migration processes are uniformly expressed. The migration characteristics that are maintained in the continuous water injection stage are identified as the borehole water injection migration patterns, and simulation results of the borehole water injection migration patterns are generated.
[0015] Secondly, the present invention provides a machine learning-based simulation system for the migration law of borehole water injection, including an identification module, which collects multi-source state data of borehole water injection during the process of borehole water injection, and identifies the data according to the collection time and spatial location of the multi-source state data of borehole water injection to form a raw state dataset of borehole water injection. The preprocessing module preprocesses the original state dataset of borehole water injection to form a borehole water injection state database, and performs feature analysis on the borehole water injection state database to obtain the hidden state feature information of groundwater migration. The analysis module performs continuous change analysis based on the characteristics of groundwater transport and migration to obtain the life cycle status information of the hydraulic channel; The module constructs a dynamic migration simulation model for borehole water injection based on the hidden state characteristics of groundwater migration and the life cycle state information of hydraulic channels. The simulation module uses a borehole water injection dynamic migration simulation model to simulate the water migration path, influence range, and migration trend during the borehole water injection process, and obtains the simulation results of borehole water injection dynamic migration. The simulation results module generates simulation results of the dynamic migration law of borehole water injection based on the simulation results of borehole water injection.
[0016] The beneficial effects of this invention are as follows: By collecting data on changes in injection pressure, injection flow rate, cumulative injection volume, hydraulic response, and geological structure during the borehole water injection process, and identifying these data in conjunction with the collection time and spatial location, a borehole water injection state database is formed. Based on this database, pressure response delay relationships and spatial migration delay relationships are extracted to analyze the continuous change process of groundwater migration latent state characteristics. Through the inherited state, newly added state, and reconstructed state change relationships in the groundwater migration latent state characteristics, the life cycle state information of the hydraulic channel is determined. Combining the groundwater migration latent state characteristics and the hydraulic channel life cycle state information, a borehole water injection dynamic migration simulation model is constructed. The water migration process is simulated through state-driven relationships and dynamic state evolution rules to obtain borehole water injection dynamic migration simulation results. Finally, the water migration path, influence range, and migration trend change law in the continuous water injection stage are extracted to generate borehole water injection migration law simulation results, realizing the simulation analysis of the dynamic evolution process of groundwater hydraulic channels and water migration law during borehole water injection. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. 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.
[0018] Figure 1 This is a flowchart of a machine learning-based simulation method for the migration patterns of water injected into boreholes.
[0019] Figure 2 This is a schematic diagram of a machine learning-based simulation system for the migration patterns of water injected into boreholes. Detailed Implementation
[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.
[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.
[0023] Reference Figures 1-2 This is one embodiment of the present invention, which provides a machine learning-based method for simulating the migration patterns of borehole water injection, comprising the following steps: S1. Collect multi-source status data of borehole water injection during the borehole water injection process, and identify the data according to the collection time and spatial location of the multi-source status data of borehole water injection to form a raw status dataset of borehole water injection.
[0024] S1.1 Collect data on changes in water injection pressure, water injection flow rate, cumulative water injection volume, hydraulic response, and geological structure correlation during the borehole water injection process.
[0025] Furthermore, multi-source state data is collected to investigate the input-response relationship and changes in the underground medium during the water injection process. Water injection pressure change data is continuously acquired through a pressure acquisition device installed in the water injection pipeline to record the state of water injection pressure changes over time during the water injection process. Water injection flow rate change data is acquired through a flow rate acquisition device to characterize the change in the amount of water entering the borehole area per unit time. Cumulative water injection volume data is obtained by accumulating the water injection flow rate change data in chronological order to describe the continuous change in the amount of water input during the water injection process.
[0026] Hydraulic response data is acquired through hydraulic response acquisition points located at different spatial positions around the borehole. By collecting information on pressure response changes, water level changes, or seepage response changes in different areas, the spatial response characteristics generated after water enters the underground medium are obtained. Geological structure correlation data is acquired through borehole geological survey data, rock strata distribution information, and spatial structure parameters to characterize the differences in geological structure corresponding to different spatial areas. The water injection input state, underground response state, and geological structure state are collected synchronously to form multi-source state data of borehole water injection, which reflects the changes in external input and underground medium response during the water injection process.
[0027] S1.2. Based on the water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data, and geological structure association data in the borehole water injection multi-source state data, record the corresponding data acquisition time as the acquisition time identifier, and generate a spatial location identifier based on the borehole depth location, spatial coordinate location, and monitoring area location corresponding to the data acquisition location.
[0028] Furthermore, based on the actual acquisition time of data such as water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data, and geological structure association data, the corresponding time information during the data acquisition process is used as the acquisition time identifier, so that each set of borehole water injection multi-source state data has a corresponding time index. Spatial location identifiers are generated according to the data acquisition locations corresponding to different data sources. For data acquired inside the borehole, spatial location identifiers are generated according to the borehole depth location corresponding to the acquisition location; for underground response data, spatial location identifiers are generated according to the spatial coordinates corresponding to the hydraulic response acquisition location; for geological structure association data, spatial location identifiers are generated according to the rock strata distribution area and the location of the monitoring area. By configuring the acquisition time identifier and spatial location identifier into the water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data, and geological structure association data respectively, data from different sources all have positioning relationships in both time and spatial dimensions.
[0029] S1.3. Based on the acquisition time identifier and the spatial location identifier, match the water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data and geological structure association data corresponding to the same acquisition time and the same spatial location to form a borehole water injection original state dataset.
[0030] Furthermore, data with the same acquisition time identifier are temporally correlated. Data on water injection pressure changes, water injection flow rate changes, cumulative water injection volume, hydraulic response, and geological structure correlation collected within the same time period are combined. Based on spatial location identifiers, data corresponding to borehole locations, underground response locations, and geological structure regions are spatially matched to maintain spatial correlation between data from different spatial regions within the same time period. Through joint matching of temporal and spatial correspondences, discretely acquired data on water injection pressure changes, water injection flow rate changes, cumulative water injection volume, hydraulic response, and geological structure correlation are organized into a borehole water injection original state dataset with temporal evolution and spatial distribution relationships.
[0031] S2. Preprocess the original state dataset of borehole water injection to form a borehole water injection state database. Perform feature analysis on the borehole water injection state database to obtain the hidden state feature information of groundwater migration.
[0032] S2.1 Based on the water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data, and geological structure association data in the original state dataset of borehole water injection, perform format unification, abnormal data correction, and missing data supplementation on different types of data to obtain the processed original state dataset of borehole water injection.
[0033] Furthermore, based on the data representation formats of different data types, data formats from different sources are converted, transforming data obtained from different acquisition methods into a data format with unified time identifiers, spatial location identifiers, and data representation rules. This allows data on changes in injection pressure, injection flow rate, cumulative injection volume, hydraulic response, and geological structure association to be stored under the same data structure. Based on the data change trends corresponding to adjacent acquisition times, it is determined whether the currently acquired data conforms to the continuous change pattern during the injection process. When data on changes in injection pressure, injection flow rate, or hydraulic response exhibits a data state that does not match the change trends of adjacent time stages, the corresponding data is marked as an abnormal state and corrected according to the data change relationships corresponding to adjacent time stages. For missing original borehole injection state data, based on the acquisition time and spatial location information corresponding to the missing location, data on changes in injection pressure, injection flow rate, cumulative injection volume, hydraulic response, and geological structure association corresponding to adjacent time stages and adjacent spatial locations are searched. The missing data is supplemented using the continuous temporal change relationship and the spatial response change relationship, resulting in a processed original borehole injection state dataset.
[0034] S2.2 Based on the processed original state dataset of borehole water injection, the time and spatial location information corresponding to the water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data and geological structure association data are organized to form a borehole water injection state database.
[0035] Furthermore, based on the acquisition time identifier in the processed original borehole water injection state dataset, the data on changes in water injection pressure, water injection flow rate, cumulative water injection volume, hydraulic response, and geological structure correlation at different time stages are arranged in a time series. This allows different types of data to be stored according to the actual occurrence sequence during the borehole water injection process. The water injection state data corresponding to the borehole location, the hydraulic response data corresponding to the underground area, and the geological structure correlation data are configured according to their spatial location relationships. This establishes a correlation between the water injection input state, hydraulic response state, and geological structure state within the same spatial area. The water injection pressure, water injection flow rate, cumulative water injection volume, hydraulic response, and geological structure correlation data, after time and spatial processing, are organized and stored according to data type, acquisition time, and spatial location, forming a borehole water injection state database that includes both time-varying and spatial distribution relationships.
[0036] S2.3. Based on the water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data, and geological structure association data in the borehole water injection status database, obtain the time nodes when the water injection pressure change data changes, and obtain the time nodes when the hydraulic response data corresponding to different spatial locations reaches the corresponding change state.
[0037] Furthermore, based on the continuous acquisition time information corresponding to the water injection pressure change data, the trend of water injection pressure change data with the water injection process is obtained. When the continuously acquired water injection pressure change data changes from the current state to a new state, the time of the state change is determined as the time node when the water injection pressure change data changes. Based on the relationship between the water injection pressure change data corresponding to adjacent acquisition times, it is determined whether the water injection pressure change data has undergone a state change. When the water injection pressure change data maintains a similar trend within the continuous acquisition time, the corresponding time period is regarded as the same pressure state. When the water injection pressure change data shows a new trend, the acquisition time of the change is determined as the time node when the water injection pressure change data changes.
[0038] Specifically, based on the spatial location identifiers corresponding to the hydraulic response data, the hydraulic response data corresponding to different spatial locations are arranged into time series. Based on the continuous state of the hydraulic response data changing over time, the time nodes when the hydraulic response data corresponding to different spatial locations reach the changed state are determined. The time when the hydraulic response data changes from a non-response state to a response state, or from the current response state to a new response state, is determined as the time node when the hydraulic response data reaches the corresponding changed state. The sets of time nodes when the injection pressure change data changes and the sets of time nodes when the hydraulic response data corresponding to different spatial locations reach the changed state are obtained respectively.
[0039] S2.4 Establish a pressure response delay relationship based on the time interval between the time nodes of the water injection pressure change data and the hydraulic response data, and establish a spatial migration delay relationship based on the time interval between the time nodes of the hydraulic response data corresponding to different spatial locations, thereby obtaining borehole water injection response delay characteristic information.
[0040] Furthermore, the time nodes of changes in water injection pressure data are matched with the time nodes of changes in hydraulic response data at corresponding spatial locations. The time nodes of changes in water injection pressure data are subtracted from the time nodes of changes in hydraulic response data to obtain the time interval required for the water injection pressure change to transmit to the corresponding spatial location and generate a hydraulic response. This time interval is used as the pressure response delay relationship. Based on the time nodes of changes in hydraulic response data at different spatial locations, the time delay relationship of water propagation from the initial spatial location to the target spatial location is obtained. This time delay relationship is used as the spatial migration delay relationship. The pressure response delay relationships corresponding to different water injection stages and the spatial migration delay relationships corresponding to different spatial locations are combined according to the water injection time sequence and spatial location relationship to form borehole water injection response delay feature information that includes water injection response time change characteristics and spatial migration time change characteristics.
[0041] S2.5. Based on the groundwater migration state change characteristics information, continuously change learning is performed on the pressure response delay relationship and spatial migration delay relationship corresponding to different water injection stages. By analyzing the change trend, change magnitude and stage change characteristics of the pressure response delay relationship and spatial migration delay relationship in the continuous water injection stage, when the change magnitude of the pressure response delay relationship and spatial migration delay relationship in the continuous water injection stage is lower than the preset change range, it is determined that the groundwater is in the initial response state.
[0042] Furthermore, based on the changing relationship between pressure response delays in adjacent water injection stages, the variation amplitude of the pressure response delay relationship during continuous water injection is calculated. Similarly, based on the changing relationship between spatial migration delays in adjacent water injection stages, the variation amplitude of the spatial migration delay relationship at different spatial locations is calculated. Taking the pressure response delay relationship and spatial migration delay relationship corresponding to each continuous water injection stage as state change objects, the characteristics of pressure response delay change and spatial migration delay change are obtained by comparing the degree of change in the delay relationship between the current water injection stage and the previous water injection stage. For the pressure response delay relationship, when the water injection pressure data changes during continuous water injection stages, the time interval between the hydraulic response data reaching the changed state remains relatively consistent. Within the range of variation, it is determined that the pressure response delay relationship has not undergone a significant state change. For the spatial migration delay relationship, when the time intervals between the hydraulic response data time nodes corresponding to different spatial locations remain similar, it is determined that the spatial migration delay relationship has not undergone a significant state change. Based on the characteristics of pressure response delay change and spatial migration delay change, the delay change state corresponding to the continuous water injection stage is jointly judged. The change amplitudes of pressure response delay relationship and spatial migration delay relationship are compared with the preset change ranges respectively. When both the change amplitudes of pressure response delay relationship and spatial migration delay relationship are within the preset change ranges, it indicates that the underground response still maintains its original propagation state, and the corresponding water injection stage is determined as the initial response state.
[0043] Definition of the first Pressure response delay relationship for each water injection stage: ; in, For the first The pressure response delay relationship corresponding to each water injection stage For the first The time points at which the data on water injection pressure changes during each water injection stage change. For the first The time points at which hydraulic response data changes during each water injection phase This is the water injection stage.
[0044] Define the spatial migration delay relationship between different spatial locations: ; in, For the first The starting spatial location of each water injection stage to target spatial location Spatial migration delay relationship, For the first The starting spatial location marker for each water injection stage Corresponding to the time points of change in hydraulic response data, For the first Target spatial location of each water injection stage Corresponding to the time points of change in hydraulic response data, The starting spatial position, The target spatial location.
[0045] Calculation method for trajectory constraint: ; in, For the first The magnitude of the delay change corresponding to each water injection stage For the first The pressure response delay relationship or spatial migration delay relationship corresponding to each water injection stage This represents the pressure response delay or spatial migration delay relationship corresponding to the next water injection stage. To prevent adjustment parameters from having a denominator of zero.
[0046] Calculation of stage change characteristics: ; in, For the first The delayed change trend characteristics corresponding to each water injection stage The number of consecutive water injection phases used for trend judgment. This is a function to determine the direction of change.
[0047] when: This indicates that the pressure response delay relationship or spatial migration delay relationship is continuously decreasing; when: This indicates an adjustment in the direction of change in the pressure response delay relationship or spatial migration delay relationship; when: This indicates a continuous increase in the pressure response delay or spatial migration delay relationship.
[0048] S2.6 When the pressure response delay relationship and spatial migration delay relationship show a continuous decreasing change characteristic during the continuous water injection stage, it is determined that the groundwater is in the state of hydraulic channel formation and expansion.
[0049] Furthermore, based on the borehole water injection response delay characteristic information, the pressure response delay relationship and spatial migration delay relationship corresponding to the continuous water injection stage are arranged in time series, and pressure response delay change sequence and spatial migration delay change sequence are established respectively. The pressure response delay relationship corresponding to the current water injection stage is compared with the pressure response delay relationship corresponding to the previous water injection stage. When the pressure response delay relationship corresponding to the subsequent water injection stage is less than the pressure response delay relationship corresponding to the previous water injection stage, the corresponding change process is recorded as a pressure response delay decreasing state. When multiple consecutive water injection stages have a pressure response delay decreasing state, the change process is determined as a continuous decreasing pressure response delay change feature.
[0050] Specifically, the spatial migration delay relationship corresponding to different spatial locations is processed in the same way. The spatial migration delay relationship corresponding to adjacent water injection stages is compared. When the time node of the hydraulic response data corresponding to the target spatial location reaches the change state gradually approaches the time node of the hydraulic response data corresponding to the starting spatial location, it is determined that the spatial migration delay relationship has decreased. When the spatial migration delay relationship between different spatial locations in the continuous water injection stage continues to decrease, the corresponding change process is determined to be a continuous decreasing change characteristic of spatial migration delay.
[0051] Based on the combined characteristics of continuously decreasing pressure response delay and continuously decreasing spatial migration delay, a continuous decrease in the pressure response delay indicates a shortening of the time required for changes in injection pressure to propagate to the underground response area. A continuous decrease in the spatial migration delay indicates a shortening of the propagation time of the hydraulic response between different spatial locations. Based on the synchronous change of these two types of delays, the groundwater migration state is determined to be transitioning from the initial response state to the hydraulic channel formation and expansion state. When the pressure response delay continuously decreases but the spatial migration delay does not change significantly, it indicates that the impact of injection is mainly concentrated in a local area, and the groundwater is determined to be in the hydraulic channel formation stage. When both the pressure response delay and the spatial migration delay show continuous decreasing changes, it indicates that the underground response area is gradually expanding, and the groundwater is determined to be in the hydraulic channel expansion stage.
[0052] S2.7 When the variation amplitude of the pressure response delay relationship and the spatial migration delay relationship gradually decreases during the continuous water injection stage and maintains the consistent variation characteristics of the continuous stage, it is determined that the groundwater is in a stable migration state.
[0053] Furthermore, based on the change process of the pressure response delay relationship corresponding to the continuous water injection stage, the change state of the pressure response delay relationship between adjacent water injection stages is obtained, and it is determined whether the pressure response delay relationship changes from a rapidly changing state to a continuously maintained state. Based on the change process of the spatial migration delay relationship corresponding to different spatial locations, the change state of the hydraulic response propagation time between different spatial regions is obtained, and it is determined whether the spatial migration delay relationship changes from an expanding change state to a continuously maintained state. When the change direction of the pressure response delay relationship remains unchanged during the continuous water injection process, but the degree of change between adjacent water injection stages gradually decreases, it indicates that the process of water injection pressure change propagating to the underground region gradually forms a stable response relationship. When the spatial migration delay relationship maintains the same change direction between different spatial locations, and no new expanding change occurs in the spatial migration relationship during the continuous water injection stage, it indicates that no new spatial adjustment has occurred in the water migration path. Based on the common change characteristics of the pressure response delay relationship and the spatial migration delay relationship, the stage of groundwater migration is determined. When both the pressure response delay relationship and the spatial migration delay relationship change from a continuously changing state to a maintained staged change state, the corresponding water injection stage is determined as a stable migration state. At this point, a relatively fixed response relationship is formed between pressure response changes, hydraulic response propagation, and spatial migration processes, and the corresponding state is regarded as the stable migration stage of groundwater.
[0054] S2.8. Based on the characteristics of the pressure response delay relationship and spatial migration delay relationship changes corresponding to different water injection stages, obtain the hidden state characteristics of groundwater migration.
[0055] Furthermore, the pressure response delay relationship changes corresponding to different water injection stages are divided into stages according to the initial response state, the formation and expansion state of hydraulic channels, and the stable migration state. The pressure response delay relationship change characteristics corresponding to each stage are correlated, and the spatial migration delay relationship change characteristics corresponding to different spatial locations are configured with the pressure response delay relationship change characteristics to form the underground response change state under different water injection stages. The state in which the pressure response delay relationship changes but the spatial migration delay relationship does not expand significantly is recorded as the local response state; the state in which the pressure response delay relationship and the spatial migration delay relationship change simultaneously are recorded as the regional migration state; and the state in which the pressure response delay relationship and the spatial migration delay relationship maintain continuous change characteristics is recorded as the stable migration state. Based on the underground response change state corresponding to different water injection stages, the pressure response delay relationship change characteristics, the spatial migration delay relationship change characteristics, and the corresponding hydraulic channel stage states are combined to form groundwater transport latent state characteristic information that can describe the changes in underground regional connectivity, changes in water body migration state, and the evolution process of hydraulic channels.
[0056] S3. Continuous change analysis is performed based on the characteristics of groundwater transport and migration to obtain the life cycle status information of the hydraulic channel.
[0057] S3.1. Based on the groundwater migration and hidden state characteristic information, establish a groundwater migration and hidden state inheritance sequence according to the continuous water injection stage, match the corresponding relationship of groundwater migration and hidden state characteristic information between adjacent water injection stages, and determine the groundwater migration and hidden state characteristic information that exists in all continuous water injection stages as the inheritance state.
[0058] Furthermore, based on the sequential water injection stages during the borehole water injection process, the groundwater migration and hidden state characteristic information corresponding to different water injection stages is organized into a time series, establishing a groundwater migration and hidden state inheritance sequence. This sequence records the changes in the connectivity, hydraulic response, and water migration states of the underground region under different water injection stages. Based on the spatial location identifiers corresponding to the groundwater migration and hidden state characteristic information, spatial matching is performed on the groundwater migration and hidden state characteristic information corresponding to the same spatial region. Finally, based on the state type corresponding to the groundwater migration and hidden state characteristic information, groundwater migration and hidden state characteristic information representing the same underground region within the same spatial region is further matched. State matching is performed on the groundwater transport and migration latent state feature information of the response state. After completing the correspondence matching between adjacent water injection stages, the retention status of the groundwater transport and migration latent state feature information in the continuous water injection stages is judged. When a certain groundwater transport and migration latent state feature information exists in both adjacent water injection stages, and the corresponding spatial location and state type maintain the correspondence, the groundwater transport and migration latent state feature information is determined to be an inherited state. If the hydraulic response state that exists in the previous water injection stage still exists in the corresponding spatial area of the subsequent water injection stage, then the groundwater transport and migration latent state feature information corresponding to the hydraulic response state is determined to be an inherited state.
[0059] S3.2. The groundwater migration and hidden state feature information that first appears in the continuous water injection stage is determined as the new state, and the groundwater migration and hidden state feature information that disappears in the continuous water injection stage is determined as the reconstructed state. The inherited feature information of the groundwater migration and hidden state is obtained based on the inherited state, the new state and the reconstructed state.
[0060] Furthermore, the newly emerging groundwater migration and occultation state characteristics in the subsequent water injection stage are compared with those in the previous water injection stage. Based on the spatial location identifier, state type identifier, and response state information corresponding to the groundwater migration and occultation state characteristics, groundwater migration and occultation state characteristics that appear in the subsequent water injection stage but do not exist in the previous water injection stage are judged. When a certain groundwater migration and occultation state characteristic first appears in a continuous water injection stage, and the corresponding spatial location and groundwater response state do not appear in the previous water injection stage, the groundwater migration and occultation state characteristic is determined to be a newly added state. If a new spatial area first appears with a hydraulic response state, then the groundwater migration and occultation state characteristic of the corresponding area is determined to be a newly added state. This process is applied to the water injection stages. The groundwater transport and migration latent state feature information in the segment is reverse-matched. Groundwater transport and migration latent state feature information that existed in the previous water injection stage but was not maintained in the subsequent water injection stage is identified as the disappearance state. When the groundwater transport and migration latent state feature information corresponding to the disappearance state is replaced by new groundwater transport and migration latent state feature information, or when the underground response state of the corresponding spatial location changes, the groundwater transport and migration latent state feature information is identified as the reconstruction state. The inherited state, the newly added state, and the reconstruction state are combined according to the continuous water injection stage to form the groundwater transport and migration latent state inherited feature information. The groundwater transport and migration latent state inherited feature information includes the maintenance relationship, generation relationship, and change relationship of the groundwater transport and migration latent state feature information in the continuous water injection process, which is used to describe the evolution of the groundwater hydraulic state with the water injection process.
[0061] S3.3. Based on the inherited feature information of groundwater migration and concealment state, the proportion of the number of groundwater migration and concealment state feature information corresponding to the inherited state in the continuous water injection stage to the total number of groundwater migration and concealment state feature information is calculated to obtain the inheritance ratio.
[0062] Furthermore, based on the state type identifier recorded in the inherited feature information of groundwater migration and displacement, the feature information of groundwater migration and displacement in the continuous water injection stage is divided into inherited state, newly added state, and reconstructed state. A set of feature information of groundwater migration and displacement corresponding to each water injection stage is obtained, and the number of groundwater migration and displacement feature information belonging to the inherited state in the set is counted. The feature information of groundwater migration and displacement is filtered according to the inherited state identifier, and the groundwater migration and displacement feature information that maintains spatial location and state type relationships between continuous water injection stages is taken as the groundwater migration and displacement feature information corresponding to the inherited state. This process is applied to all groundwater migration and displacement feature information in the current water injection stage. The characteristics of groundwater transport and migration are statistically analyzed to obtain the total number of groundwater transport and migration characteristics corresponding to the current water injection stage. Based on the correspondence between the number of groundwater transport and migration characteristics corresponding to the inherited state and the total number of groundwater transport and migration characteristics, the inheritance ratio is obtained. This ratio is used to characterize the degree of change in the continuity of groundwater transport and migration characteristics during continuous water injection stages. If the groundwater transport and migration characteristics formed in the previous stage still maintain the corresponding relationship in the subsequent stage, then the groundwater transport and migration characteristics are statistically counted as inherited states. When multiple continuous water injection stages maintain the same continuity relationship, the corresponding number of inherited states is continuously retained, forming a corresponding inheritance ratio change process.
[0063] S3.4. Calculate the proportion of groundwater migration latent state feature information corresponding to the reconstructed state during the continuous water injection stage to the total number of groundwater migration latent state feature information, and obtain the reconstruction ratio.
[0064] Furthermore, based on the reconstruction state identifier in the inherited feature information of groundwater migration and occultation, the feature information of groundwater migration and occultation that underwent state replacement or state disappearance during continuous water injection stages is screened. Groundwater migration and occultation feature information that existed in the previous water injection stage but did not maintain the same spatial location and state type relationship in the subsequent water injection stage is identified as the groundwater migration and occultation feature information corresponding to the reconstruction state. When the original groundwater migration and occultation feature information disappears and new groundwater migration and occultation feature information is generated in the corresponding spatial area, the state change between the preceding and following stages is recorded as a reconstruction state. The number of groundwater migration and occultation feature information corresponding to the reconstruction states in continuous water injection stages is counted, and the total number of groundwater migration and occultation feature information for the corresponding water injection stage is obtained. By statistically analyzing the correspondence between the number of reconstruction states and the total number of groundwater migration and occultation feature information, the reconstruction ratio is obtained.
[0065] S3.5 Determine the characteristic information of the hydraulic channel evolution stage based on the changes in the inheritance ratio and reconstruction ratio during the continuous water injection stage.
[0066] Furthermore, inheritance ratio change sequences and reconstruction ratio change sequences are established according to the time sequence of continuous water injection stages. By combining the changes in the two ratio sequences in different water injection stages, the hydraulic channel state change process corresponding to the groundwater migration and concealment state characteristic information is determined. Based on the changing directions of the inheritance ratio and reconstruction ratio in the continuous water injection stages, the hydraulic channel change types corresponding to the groundwater migration and concealment state characteristic information are classified. When the inheritance ratio maintains a continuous increasing state and the reconstruction ratio does not show a synchronous increasing change, the corresponding stage is determined as the continuous development stage of the hydraulic channel; when both the inheritance ratio and reconstruction ratio change alternately, the hydraulic channel state change is classified as follows: The current stage is defined as the hydraulic channel structure adjustment stage. When the reconstruction ratio continues to increase and the inheritance ratio decreases accordingly, the corresponding stage is defined as the hydraulic channel renewal stage. Based on the changes in the inheritance ratio and reconstruction ratio corresponding to different water injection stages, the hydraulic channel change types corresponding to each water injection stage are combined in chronological order to form hydraulic channel evolution stage characteristic information reflecting the process of hydraulic channel evolution from state maintenance to state expansion and state adjustment. The hydraulic channel evolution stage characteristic information includes the state type and state transition relationship corresponding to different stages during continuous water injection, which is used to describe the stage process of changes in underground hydraulic channels with the water injection process.
[0067] S3.6. Based on the characteristic information of the hydraulic channel evolution stage, the inheritance ratio and reconstruction ratio corresponding to the continuous water injection stage are continuously analyzed. When the inheritance ratio increases in the continuous water injection stage, it is determined that the hydraulic channel is in a continuous development process.
[0068] Furthermore, based on the characteristic information of the hydraulic channel evolution stages, the hydraulic channel state type corresponding to each water injection stage is determined, and the inheritance ratio and reconstruction ratio in the corresponding stages are configured in association. The direction of change of the inheritance ratio in continuous water injection stages is judged, and the corresponding relationship of the inheritance ratios of adjacent water injection stages is compared. When the inheritance ratio in the later water injection stage is higher than that in the previous water injection stage, and multiple consecutive water injection stages maintain the same direction of change, it is determined that the inheritance ratio is in an increasing state. Combined with the change state of the reconstruction ratio, the continuous development process of the hydraulic channel is determined. When the inheritance ratio continues to increase, it indicates that the groundwater migration latent state characteristic information formed in the previous stage continues to be maintained in the subsequent water injection process. At the same time, when the reconstruction ratio does not form a continuously increasing state, it indicates that the new underground state replacement process has not occupied the main change trend. Based on the process of increasing inheritance ratio and changing reconstruction ratio, the corresponding water injection stage is determined as the continuous development process of the hydraulic channel.
[0069] S3.7 When the inheritance ratio and reconstruction ratio alternate during the continuous water injection stage, it is determined that the hydraulic channel is in the adjustment process.
[0070] Furthermore, based on the relationship between the inheritance and reconstruction ratios corresponding to different water injection stages, a sequence of hydraulic channel state changes is established. The maintenance and replacement changes of groundwater migration and concealment state characteristics during continuous water injection are recorded accordingly. The directions of inheritance and reconstruction ratio changes between adjacent water injection stages are matched. When the inheritance ratio increases in a certain continuous water injection stage, and the reconstruction ratio increases in the next continuous water injection stage, the corresponding change process is recorded as a state alternation process. When the inheritance ratio resumes its change after the reconstruction ratio increases, the groundwater migration and concealment state characteristics are recorded between the maintenance and replacement states. The conversion process is determined by the alternating changes in the inheritance ratio and the reconstruction ratio to determine the hydraulic channel adjustment state. When the inheritance ratio and the reconstruction ratio do not have a single continuous change trend in the continuous water injection stage, and they alternate between different water injection stages, it indicates that the groundwater migration and hidden state characteristics information exist simultaneously in a process of maintenance and replacement. The corresponding stage is determined as the hydraulic channel adjustment process. The groundwater migration and hidden state characteristics information formed in the previous stage still exists. New groundwater migration and hidden state characteristics information are generated in other spatial areas. Then, the corresponding stage shows that the inheritance state and the reconstruction state change together. The change process is determined as the hydraulic channel adjustment process.
[0071] S3.8 When the reconstruction ratio increases during the continuous water injection stage, it is determined that the hydraulic channel is in the renewal process. Based on the hydraulic channel evolution stage characteristic information corresponding to different water injection stages, the life cycle status information of the hydraulic channel is obtained.
[0072] Furthermore, based on the reconstruction state identifier in the inherited feature information of groundwater migration and concealment status, the change process of the number of reconstruction states corresponding to different water injection stages is obtained, and the change process of the number of reconstruction states is correlated with the change process of all groundwater migration and concealment status feature information. The direction of the change in the reconstruction ratio between continuous water injection stages is judged. When the reconstruction ratio corresponding to the later water injection stage is higher than that corresponding to the previous water injection stage, and the reconstruction ratio continues to increase in the continuous water injection stages, it is determined that the groundwater migration and concealment status feature information has undergone continuous replacement change. When the original groundwater migration and concealment status feature information gradually decreases during the continuous water injection process, the new groundwater migration and concealment status feature information does not... When the same spatial region continues to generate, the corresponding state change is determined as the hydraulic channel renewal process. The groundwater migration and latent state characteristics of the original hydraulic response area disappear, and at the same time, a new hydraulic response state is generated in the new spatial region. Then, it is determined that the hydraulic channel has been renewed. Based on the hydraulic channel's continuous development process, hydraulic channel adjustment process, and hydraulic channel renewal process, the hydraulic channel evolution stage characteristics are combined in the order of continuous water injection stages to form the hydraulic channel life cycle state information. The hydraulic channel life cycle state information records the hydraulic channel state change process corresponding to different water injection stages, including the conversion relationship between hydraulic channel state maintenance, state adjustment, and state renewal.
[0073] S4. Based on the hidden state characteristics of groundwater migration and the life cycle state information of hydraulic channels, a dynamic migration simulation model for borehole water injection is constructed.
[0074] S4.1 Based on the groundwater migration and concealment state characteristic information, extract the state change process between adjacent groundwater migration and concealment state characteristic information according to the continuous water injection stage, count the water injection conditions corresponding to the transformation of each groundwater migration and concealment state characteristic information to other groundwater migration and concealment state characteristic information, and group the repeated state change processes under the same water injection conditions into the same state driving relationship to obtain groundwater migration and concealment state driving relationship information.
[0075] Furthermore, the groundwater migration and occultation status characteristics of the current water injection stage are matched with those of the previous water injection stage. Based on the spatial location identifier, state type information, and response state information corresponding to the groundwater migration and occultation status characteristics, the direction of change of the groundwater migration and occultation status characteristics between the two water injection stages is determined. When the groundwater migration and occultation status characteristics existing in the previous water injection stage maintain the same spatial location and state type in the next water injection stage, the correspondence is recorded as a state maintenance process. When the groundwater migration and occultation status characteristics of the previous water injection stage change to groundwater migration and occultation status characteristics of another spatial location or another state type, the corresponding change is recorded as a state transition process. For example, when a local response state changes to a regional migration state, the state change process is recorded as a state transition between the groundwater migration and occultation status characteristics.
[0076] Specifically, after obtaining the state transition processes corresponding to multiple consecutive water injection stages, the water injection conditions corresponding to each state transition process are associated and recorded. These water injection conditions consist of water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, and corresponding water injection stage information at the time of the state transition. When the groundwater migration latent state characteristic information changes from a first state to a second state, the changes in water injection pressure change data, water injection flow rate change data, and cumulative water injection volume data corresponding to the transition stage are recorded, forming the water injection condition information corresponding to the state transition. Based on the state transition process and the corresponding water injection condition information, the recurring state transitions in the continuous water injection stages are analyzed. The process is merged. When the same direction of change of groundwater migration latent state characteristics exists in different water injection stages, and the corresponding water injection conditions are in the same change type, multiple state change processes are merged into the same state driving relationship. When the same groundwater migration latent state characteristics change from local response state to regional migration state in multiple water injection stages, and all correspond to water injection conditions of increased water injection pressure and expanded hydraulic response range, the corresponding state change processes are merged into the same state driving relationship. The discrete state change processes between groundwater migration latent state characteristics are converted into state driving relationships with water injection condition constraints, thus obtaining groundwater migration latent state driving relationship information.
[0077] S4.2. Based on the groundwater migration and concealment state driving relationship information and the hydraulic channel life cycle state information, count the frequency of occurrence of groundwater migration and concealment state driving relationships corresponding to different hydraulic channel life cycle state information. Configure the groundwater migration and concealment state driving relationships with a frequency higher than a preset frequency threshold as the state evolution rules under the corresponding hydraulic channel life cycle state information. Combine the state evolution rules corresponding to different hydraulic channel life cycle state information to obtain a dynamic state evolution rule set.
[0078] Furthermore, the information on the driving relationships of groundwater migration and occultation in the water injection stage corresponding to the continuous development state of the hydraulic channel is categorized into the continuous development state set; the information on the driving relationships of groundwater migration and occultation in the water injection stage corresponding to the adjustment state of the hydraulic channel is categorized into the adjustment state set; and the information on the driving relationships of groundwater migration and occultation in the water injection stage corresponding to the renewal state of the hydraulic channel is categorized into the renewal state set. Frequency statistics are performed on the information on the driving relationships of groundwater migration and occultation in the sets corresponding to different hydraulic channel life cycle states. The number of times each driving relationship of groundwater migration and occultation occurs in the same life cycle state is counted, and the occurrence ratio of the driving relationship of groundwater migration and occultation is determined based on the total number of state change processes.
[0079] The preset frequency threshold is used to filter state-driven relationships with stable and recurring characteristics. The frequency threshold is determined based on the distribution of groundwater migration-induced state-driven relationships during continuous water injection. A frequency distribution sequence is formed based on the occurrence frequency of all groundwater migration-induced state-driven relationships. Groundwater migration-induced state-driven relationships with high frequency and within the main distribution range are identified as valid state-driven relationships. Groundwater migration-induced state-driven relationships with low occurrence frequency and unable to form continuous change relationships are filtered out. Groundwater migration-induced state-driven relationships exceeding the preset frequency threshold are configured as state evolution rules under the corresponding hydraulic channel lifecycle state information. If, during the continuous development of the hydraulic channel, multiple groundwater migration-induced state-driven relationships show a transition from a local response state to a regional migration state, then these groundwater migration-induced state-driven relationships are configured as state evolution rules corresponding to the continuous development state. The state evolution rules corresponding to different hydraulic channel lifecycle state information are combined according to the lifecycle stage sequence to form a dynamic state evolution rule set.
[0080] S4.3. Based on the dynamic state evolution rule set, the groundwater migration latent state feature information is used as the initial state input, and the dynamic state evolution rule set is used as the basis for updating the groundwater migration latent state. The groundwater migration latent state feature information after each water injection stage is recursively updated, and a borehole water injection dynamic migration simulation model is constructed according to the recursive result of the groundwater migration latent state feature information.
[0081] Furthermore, the groundwater transport and migration hidden state characteristic information includes the underground region connectivity change state, hydraulic response change state, and water body migration trend state, which are used to describe the underground response state corresponding to the initial stage of water injection. Based on the hydraulic channel life cycle state information corresponding to the current water injection stage, the corresponding state evolution rule is matched from the dynamic state evolution rule set. Based on the hydraulic channel life cycle state identifier corresponding to the current water injection stage, the state evolution rule under the corresponding life cycle stage is invoked, and the groundwater transport and migration hidden state characteristic information of the current stage is used as the state information input before the state update. Based on the state type corresponding to the current groundwater transport and migration hidden state characteristic information and the water injection condition information corresponding to the current water injection stage, the state transition relationship corresponding to the dynamic state evolution rule set is matched. When a corresponding state driving relationship exists, the current groundwater transport and migration hidden state characteristic information is converted into the groundwater transport and migration hidden state characteristic information corresponding to the next water injection stage; when no corresponding state driving relationship exists, the current groundwater transport and migration hidden state characteristic information is kept unchanged.
[0082] Specifically, when the current water injection stage corresponds to a continuously developing hydraulic channel, and the dynamic state evolution rule set contains state evolution rules that transition from a local response state to a regional migration state, the groundwater migration latent state characteristic information is updated according to the corresponding state evolution rules, so that the next water injection stage obtains a new underground regional response state. According to the sequence of continuous water injection stages, the groundwater migration latent state characteristic information after the completion of each water injection stage is recursively updated to form a continuous state sequence in which the groundwater migration latent state characteristic information changes with the water injection process. Based on the transformation relationship of the groundwater migration latent state characteristic information in the continuous state sequence, the groundwater migration latent state characteristic information, the hydraulic channel life cycle state information, and the dynamic state evolution rule set are associated and configured to construct a borehole water injection dynamic migration simulation model. The borehole water injection dynamic migration simulation model includes the initial underground state input relationship, the life cycle state matching relationship, and the groundwater migration latent state recursive update relationship, which is used to continuously update the groundwater migration state according to the changes in the water injection stage.
[0083] S5. The dynamic migration simulation model of borehole water injection is used to simulate the water migration path, influence range and migration trend during the borehole water injection process, and the simulation results of the dynamic migration of borehole water injection are obtained.
[0084] S5.1 Based on the life cycle status information of the hydraulic channel, establish corresponding migration rules for different migration stages in the borehole water injection dynamic migration simulation model, and match different life cycle stages corresponding to the life cycle status information of the hydraulic channel with different water body migration modes to obtain water body dynamic migration rule information.
[0085] Furthermore, based on the stage identifiers in the life cycle status information of the hydraulic channel, the continuous water injection process is divided into different migration stages corresponding to the hydraulic channel status, including the continuous development stage, the adjustment stage, and the renewal stage of the hydraulic channel. Corresponding water migration rules are established, and the water migration mode corresponding to different life cycle stages is determined based on the regional connectivity change status, hydraulic response change status, and water migration trend status in the groundwater migration latent state characteristic information.
[0086] Specifically, when the hydraulic channel is in the continuous development stage, based on the groundwater migration latent state characteristic information where the inherited state is the main change feature, water body migration rules are configured to maintain the original migration direction and gradually expand the migration range; when the hydraulic channel is in the adjustment stage, based on the groundwater migration latent state characteristic information where the inherited state and reconstruction state alternate, water body migration rules are configured to adjust the migration direction and rematch the migration path; when the hydraulic channel is in the update stage, based on the corresponding groundwater migration latent state characteristic information added in the reconstruction state, water body migration rules are configured to replace the migration path and form a new migration area. The water body migration rules corresponding to the life cycle state information of different hydraulic channels are associated and stored according to the life cycle stage, so that each hydraulic channel life cycle state information corresponds to a water body migration mode, forming dynamic water body migration rule information.
[0087] S5.2. Based on the dynamic migration rules of water bodies, the water migration process in the borehole water injection dynamic migration simulation model is controlled by stage switching. When the life cycle status information of the hydraulic channel changes in a stage, the corresponding dynamic migration rules of water bodies are switched synchronously to update the migration direction, migration expansion mode and migration advancement process of water bodies, so as to obtain dynamic migration process information.
[0088] Furthermore, during the operation of the borehole water injection dynamic migration simulation model, the life cycle state information of the hydraulic channel corresponding to the current stage is obtained according to the continuous water injection stage, and compared with the life cycle state information of the hydraulic channel corresponding to the previous water injection stage. When the life cycle state corresponding to the continuous water injection stage is consistent, the water body dynamic migration rule information corresponding to the current stage continues to be called to continuously update the water migration process. When the life cycle state information of the hydraulic channel changes, the corresponding water body dynamic migration rule information is re-matched according to the changed life cycle state information. When the hydraulic channel transitions from the continuous development stage to the adjustment stage, the original extended water body migration rule is switched off. The water migration rules are changed to path adjustment type. When the hydraulic channel transitions from the adjustment phase to the update phase, the path adjustment type water migration rules are switched to migration path replacement type water migration rules. Based on the switched water dynamic migration rule information, the water migration direction, migration expansion mode, and migration advancement process are updated. The water migration direction is determined based on the groundwater migration latent state driving relationship corresponding to the current life cycle stage; the migration expansion mode is determined based on the change of spatial migration delay relationship; and the migration advancement process is determined based on the recursive result of the groundwater migration latent state in the dynamic state evolution rule set, thus obtaining dynamic migration process information that is synchronously adjusted with the life cycle state change of the hydraulic channel.
[0089] S5.3 Based on the dynamic migration process information, the borehole water injection dynamic migration simulation model is used to continuously generate the water migration path, influence range and migration trend corresponding to different water injection stages, and obtain the borehole water injection dynamic migration simulation results.
[0090] Furthermore, based on the water migration direction and migration progress status in the dynamic migration process information, the water migration path corresponding to the current water injection stage is determined. The directional relationship of water propagation from the current spatial location to the next spatial location is determined based on the recursive results of the groundwater migration latent state. The life cycle status information of the hydraulic channel is used to determine whether the migration path is maintained, expanded, or replaced. The water body influence range is generated based on the migration expansion mode in the dynamic migration process information. The response propagation relationship between different spatial regions is determined based on the spatial migration delay relationship change process. The influence area expansion range is determined based on the water body dynamic migration rule information. The water body migration trend is generated based on the water body migration path and influence range change process corresponding to the continuous water injection stages. The changes in migration direction, influence area, and hydraulic channel status corresponding to different water injection stages are continuously correlated to form water body migration trend information. The water body migration path, influence range, and migration trend corresponding to different water injection stages are combined to form the borehole water injection dynamic migration simulation results.
[0091] S6. Based on the dynamic migration simulation results of borehole water injection, generate simulation results of borehole water injection migration law.
[0092] S6.1. Based on the dynamic migration simulation results of borehole water injection, extract the water migration path change process, influence range change process, and migration trend change process corresponding to the continuous water injection stage. Classify the repeated change processes in the continuous water injection stage to obtain migration evolution model information.
[0093] Furthermore, the water migration paths corresponding to adjacent water injection stages are correlated according to their spatial location. The starting position of the path, the direction of path extension, and the area covered by the path are used as path change characteristics to determine the change process of the water migration path in continuous water injection stages. When the water migration path of the later water injection stage maintains the same propagation direction and spatial extension relationship as the previous water injection stage, the corresponding change process is recorded as the path maintenance state. When the direction of the water migration path is adjusted or a new migration area is added, the corresponding change process is recorded as the path change state. Based on the influence range information in the borehole water injection dynamic migration simulation results, the change process of the water influence area corresponding to different water injection stages is determined. The boundary position, coverage area, and spatial expansion direction of the influence area corresponding to the continuous water injection stage are analyzed to determine the process of the influence range changing from the maintenance state to the expansion state, adjustment state, or contraction state.
[0094] Specifically, based on the changing trends of water body migration direction, migration speed, and spatial propulsion direction in continuous stages, a migration trend change process is formed. The water body migration path change process, influence range change process, and migration trend change process are combined to form the migration state change characteristics corresponding to different water injection stages. Based on the repeated change characteristics in multiple continuous water injection stages, they are classified. When different water injection stages have the same path change relationship, the same influence range change relationship, and the same migration trend change relationship, the corresponding change process is divided into the same migration evolution mode, and migration evolution mode information is obtained.
[0095] S6.2. Based on the transport evolution pattern information, continuously track the connection relationship between different transport evolution pattern information, combine the transport evolution pattern information that can be continuously connected into a complete transport process, divide the transport evolution pattern information that has been interrupted or replaced into different transport processes, and obtain transport law association information.
[0096] Furthermore, the characteristics of water migration path changes, influence range changes, and migration trend changes in the previous stage's migration evolution model information are compared with the corresponding characteristics in the subsequent stage's migration evolution model information. When the subsequent stage's migration evolution model information can inherit the path change direction, influence range expansion relationship, and migration trend change direction of the previous stage, the two migration evolution model information are determined to be in a continuous connection relationship. If the previous stage shows a continuous expansion of the migration range, and the subsequent stage continues to maintain the same expansion direction, then the two migration evolution model information are determined to belong to the same continuous migration process. When there are changes in path direction, replacement of affected area, or change in migration trend compared to the previous stage, the corresponding migration evolution model information is determined to be discontinuous. When the water body migration path changes from one spatial region to another and the original area of influence is no longer maintained, the corresponding change process is determined to be a migration process replacement. Based on the continuous connection relationship between migration evolution model information, multiple migration evolution model information with continuous evolution relationship are combined to form a complete migration process. Migration evolution model information with interruption or replacement relationship is divided into different migration processes, and migration law association information is formed according to the conversion relationship between each migration process.
[0097] S6.3 Based on the information related to the migration law, the water migration path, influence range and migration trend corresponding to different migration processes are uniformly expressed. The migration characteristics that are maintained in the continuous water injection stage are determined as the borehole water injection migration law, and the simulation results of the borehole water injection migration law are generated.
[0098] Furthermore, the water migration paths corresponding to multiple stages are connected in chronological order to form a continuous migration path change process; the influence range change processes corresponding to different stages are correlated to form an influence area evolution process; the migration trend change processes corresponding to different stages are integrated to form an overall migration trend change process; the water migration path change relationship, influence range change relationship, and migration trend change relationship that exist in different complete migration processes are correspondingly organized; when multiple migration processes maintain the same migration direction, the same expansion mode, and the same trend change relationship, the corresponding change characteristics are determined as stable migration characteristics; the water migration law in the borehole water injection process is expressed based on the stable migration characteristics, forming a borehole water injection migration law simulation result that includes the migration path change law, influence range change law, and migration trend change law.
[0099] This embodiment also provides a machine learning-based simulation system for borehole water injection migration patterns, including: The identification module collects multi-source status data of borehole water injection during the borehole water injection process, and identifies the data according to the collection time and spatial location of the multi-source status data of borehole water injection to form a raw status dataset of borehole water injection. The preprocessing module preprocesses the original state dataset of borehole water injection to form a borehole water injection state database, and performs feature analysis on the borehole water injection state database to obtain the hidden state feature information of groundwater migration. The analysis module performs continuous change analysis based on the characteristics of groundwater transport and migration to obtain the life cycle status information of the hydraulic channel; The module constructs a dynamic migration simulation model for borehole water injection based on the hidden state characteristics of groundwater migration and the life cycle state information of hydraulic channels. The simulation module uses a borehole water injection dynamic migration simulation model to simulate the water migration path, influence range, and migration trend during the borehole water injection process, and obtains the simulation results of borehole water injection dynamic migration. The simulation results module generates simulation results of the dynamic migration law of borehole water injection based on the simulation results of borehole water injection.
[0100] In summary, this invention collects data on changes in injection pressure, flow rate, cumulative injection volume, hydraulic response, and geological structure during the borehole water injection process. These data are then labeled with the collected time and spatial location to form a borehole water injection state database. Based on this database, pressure response delay and spatial migration delay relationships are extracted to analyze the continuous change process of the hidden state characteristics of groundwater migration. By analyzing the inherited, newly added, and reconstructed state changes within the hidden state characteristics of groundwater migration, the life cycle state information of the hydraulic channel is determined. A dynamic migration simulation model for borehole water injection is constructed by combining the hidden state characteristics of groundwater migration with the life cycle state information of the hydraulic channel. The water migration process is simulated using state-driven relationships and dynamic state evolution rules to obtain the simulation results. Finally, the migration path, influence range, and migration trend change patterns of the water body during the continuous injection stage are extracted to generate simulation results of the borehole water injection migration patterns. This enables the simulation and analysis of the dynamic evolution process of the groundwater hydraulic channel and the migration patterns of the water body during borehole water injection.
[0101] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A method for simulating the migration patterns of borehole water injection based on machine learning, characterized in that: include, Collect multi-source state data of borehole water injection during the borehole water injection process, and identify the data according to the collection time and spatial location of the multi-source state data of borehole water injection to form a raw state dataset of borehole water injection. The original state dataset of borehole water injection is preprocessed to form a borehole water injection state database. Feature analysis is then performed on the borehole water injection state database to obtain the hidden state feature information of groundwater migration. Continuous change analysis is performed based on the characteristics of groundwater transport and migration to obtain the life cycle status information of hydraulic channels; Based on the hidden state characteristics of groundwater migration and the life cycle state information of hydraulic channels, a dynamic migration simulation model of borehole water injection is constructed. The dynamic migration simulation model of borehole water injection was used to simulate the water migration path, influence range and migration trend during the borehole water injection process, and the simulation results of borehole water injection dynamic migration were obtained. Based on the simulation results of borehole water injection dynamic migration, simulation results of borehole water injection migration law are generated. 2.The machine learning based simulation method of waterflood sweep efficiency according to claim 1, wherein, The multi-source status data of borehole water injection during the collection process includes: Data on changes in water injection pressure, water injection flow rate, cumulative water injection volume, hydraulic response, and geological structure correlation were collected during the borehole water injection process. 3.The method of claim 2, wherein, The step of identifying the original state dataset of borehole water injection based on the acquisition time and spatial location corresponding to the multi-source state data of borehole water injection includes: Based on the water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data, and geological structure association data in the borehole water injection multi-source state data, the corresponding data acquisition time is recorded as the acquisition time identifier, and a spatial location identifier is generated based on the borehole depth location, spatial coordinate location, and monitoring area location corresponding to the data acquisition location. Based on the acquisition time identifier and the spatial location identifier, the data on changes in water injection pressure, changes in water injection flow rate, cumulative water injection volume, hydraulic response data, and geological structure correlation data corresponding to the same acquisition time and spatial location are matched to form a dataset of the original state of borehole water injection. 4.The method of claim 3, wherein, The preprocessing of the original borehole water injection state dataset to form a borehole water injection state database includes: Based on the water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data, and geological structure association data in the original state dataset of borehole water injection, the different types of data are formatted, abnormal data is corrected, and missing data is supplemented to obtain the processed original state dataset of borehole water injection. Based on the processed raw data set of borehole water injection, the time and spatial location information corresponding to the data on changes in water injection pressure, changes in water injection flow rate, cumulative water injection volume, hydraulic response, and geological structure are organized to form a borehole water injection status database.
5. The method for simulating borehole water injection migration patterns based on machine learning as described in claim 4, characterized in that, The feature analysis of the borehole water injection status database to obtain groundwater migration and latent state feature information includes: Based on the water injection pressure change data, water injection flow rate change data, cumulative water injection volume data, hydraulic response data, and geological structure correlation data in the borehole water injection status database, the time nodes when the water injection pressure change data changes are obtained, and the time nodes when the hydraulic response data corresponding to different spatial locations reaches the corresponding change state are obtained. The pressure response delay relationship is established based on the time interval between the time nodes of the water injection pressure change data and the hydraulic response data, and the spatial migration delay relationship is established based on the time interval between the time nodes of the hydraulic response data corresponding to different spatial locations, so as to obtain the borehole water injection response delay characteristic information. Based on the groundwater migration state change characteristics, the pressure response delay relationship and spatial migration delay relationship corresponding to different water injection stages are continuously changed and learned. By analyzing the change trend, change magnitude and stage change characteristics of the pressure response delay relationship and spatial migration delay relationship in the continuous water injection stage, when the change magnitude of the pressure response delay relationship and spatial migration delay relationship in the continuous water injection stage is lower than the preset change range, the groundwater is determined to be in the initial response state. When the pressure response delay relationship and spatial migration delay relationship show a continuous decreasing change during the continuous water injection stage, it is determined that the groundwater is in the state of hydraulic channel formation and expansion. When the variation amplitudes of pressure response delay relationship and spatial migration delay relationship gradually decrease during the continuous water injection stage and maintain consistent variation characteristics in the continuous stage, it is determined that the groundwater is in a stable migration state. Based on the changing characteristics of pressure response delay and spatial migration delay at different water injection stages, information on the latent state of groundwater migration is obtained.
6. The method for simulating borehole water injection migration patterns based on machine learning as described in claim 5, characterized in that, The continuous change analysis based on the groundwater transport and concealment status characteristics to obtain the life cycle status information of the hydraulic channel includes: Based on the groundwater migration and hidden state feature information, a groundwater migration and hidden state inheritance sequence is established according to the continuous water injection stage. The correspondence relationship of the groundwater migration and hidden state feature information between adjacent water injection stages is matched, and the groundwater migration and hidden state feature information that exists in all continuous water injection stages is determined as the inheritance state. The groundwater migration and latent state feature information that first appears during the continuous water injection stage is identified as the new state, and the groundwater migration and latent state feature information that disappears during the continuous water injection stage is identified as the reconstructed state. The inherited feature information of the groundwater migration and latent state is obtained based on the inherited state, the new state, and the reconstructed state. Based on the inherited feature information of groundwater migration hidden state, the proportion of the number of groundwater migration hidden state feature information corresponding to the inherited state in the continuous water injection stage to the total number of groundwater migration hidden state feature information is calculated to obtain the inheritance ratio. The reconstruction ratio is obtained by statistically analyzing the proportion of groundwater migration latent state feature information corresponding to the reconstructed state during the continuous water injection stage to the total number of groundwater migration latent state feature information. The characteristic information of the hydraulic channel evolution stage is determined based on the changes in the inheritance ratio and reconstruction ratio during the continuous water injection stage; Based on the characteristic information of the hydraulic channel evolution stage, the inheritance ratio and reconstruction ratio corresponding to the continuous water injection stage are continuously analyzed. When the inheritance ratio increases in the continuous water injection stage, it is determined that the hydraulic channel is in a continuous development process. When the inheritance ratio and reconstruction ratio alternate during the continuous water injection phase, it is determined that the hydraulic channel is in the adjustment process. When the reconstruction ratio increases during the continuous water injection phase, it is determined that the hydraulic channel is in the renewal process. Based on the characteristic information of the hydraulic channel evolution stage corresponding to different water injection phases, the life cycle status information of the hydraulic channel is obtained.
7. The method for simulating borehole water injection migration patterns based on machine learning as described in claim 6, characterized in that, The dynamic migration simulation model for borehole water injection, based on the latent state characteristics of groundwater migration and the life cycle state information of hydraulic channels, includes: Based on the groundwater migration and hidden state characteristic information, the state change process between adjacent groundwater migration and hidden state characteristic information is extracted according to the continuous water injection stage. The water injection conditions corresponding to the transformation of each groundwater migration and hidden state characteristic information to other groundwater migration and hidden state characteristic information are statistically analyzed. The repeated state change processes under the same water injection conditions are grouped into the same state driving relationship to obtain the groundwater migration and hidden state driving relationship information. Based on the groundwater transport and migration state driving relationship information and the hydraulic channel life cycle state information, the frequency of occurrence of groundwater transport and migration state driving relationships corresponding to different hydraulic channel life cycle state information is counted. Groundwater transport and migration state driving relationships with a frequency higher than a preset frequency threshold are configured as state evolution rules under the corresponding hydraulic channel life cycle state information. The state evolution rules corresponding to different hydraulic channel life cycle state information are combined to obtain a dynamic state evolution rule set. Based on the dynamic state evolution rule set, the hidden state characteristics of groundwater migration are used as the initial state input, and the dynamic state evolution rule set is used as the basis for updating the hidden state of groundwater migration. The hidden state characteristics of groundwater migration after each water injection stage are recursively updated, and a borehole water injection dynamic migration simulation model is constructed according to the recursive results of the hidden state characteristics of groundwater migration.
8. The method for simulating borehole water injection migration patterns based on machine learning as described in claim 7, characterized in that, The simulation model of dynamic migration during borehole water injection is used to simulate the water migration path, influence range, and migration trend during the borehole water injection process. The simulation results of dynamic migration during borehole water injection include: Based on the life cycle status information of the hydraulic channel, corresponding migration rules are established for different migration stages in the dynamic migration simulation model of borehole water injection. Different water body migration modes are matched with different life cycle stages corresponding to the life cycle status information of the hydraulic channel to obtain water body dynamic migration rule information. Based on the dynamic migration rules of water bodies, the water migration process in the borehole water injection dynamic migration simulation model is controlled by stage switching. When the life cycle status information of the hydraulic channel changes in a stage, the corresponding dynamic migration rules of water bodies are switched synchronously to update the water migration direction, migration expansion mode and migration advancement process, thereby obtaining dynamic migration process information. Based on the dynamic migration process information, the borehole water injection dynamic migration simulation model is used to continuously generate the water migration path, influence range and migration trend corresponding to different water injection stages, and obtain the borehole water injection dynamic migration simulation results.
9. The method for simulating borehole water injection migration patterns based on machine learning as described in claim 8, characterized in that, The generation of borehole water injection migration law simulation results based on the borehole water injection dynamic migration simulation results includes: Based on the dynamic migration simulation results of borehole water injection, the changes in water migration path, influence range, and migration trend corresponding to the continuous water injection stage are extracted. The repeated changes in the continuous water injection stage are classified to obtain migration evolution pattern information. Based on the information of transport evolution patterns, the connection between different transport evolution patterns is continuously tracked. Transport evolution patterns that can be continuously connected are combined into a complete transport process. Transport evolution patterns that are interrupted or replaced are divided into different transport processes to obtain transport pattern association information. Based on the information related to the migration patterns, the water migration paths, influence ranges, and migration trends corresponding to different migration processes are uniformly expressed. The migration characteristics that are maintained in the continuous water injection stage are identified as the borehole water injection migration patterns, and simulation results of the borehole water injection migration patterns are generated.
10. A machine learning-based simulation system for borehole water injection migration patterns, based on the machine learning-based simulation method for borehole water injection migration patterns according to any one of claims 1 to 9, characterized in that, include, The identification module collects multi-source status data of borehole water injection during the borehole water injection process, and identifies the data according to the collection time and spatial location of the multi-source status data of borehole water injection to form a raw status dataset of borehole water injection. The preprocessing module preprocesses the original state dataset of borehole water injection to form a borehole water injection state database, and performs feature analysis on the borehole water injection state database to obtain the hidden state feature information of groundwater migration. The analysis module performs continuous change analysis based on the characteristics of groundwater transport and migration to obtain the life cycle status information of the hydraulic channel; The module constructs a dynamic migration simulation model for borehole water injection based on the hidden state characteristics of groundwater migration and the life cycle state information of hydraulic channels. The simulation module uses a borehole water injection dynamic migration simulation model to simulate the water migration path, influence range, and migration trend during the borehole water injection process, and obtains the simulation results of borehole water injection dynamic migration. The simulation results module generates simulation results of the dynamic migration law of borehole water injection based on the simulation results of borehole water injection.
