Storage potential prediction method and system based on mine water depth geological storage technology

Abstract

The invention discloses a storage potential prediction method and system based on a mine water depth geological storage technology, and relates to the technical field of mine water treatment.The method comprises the steps that monitoring data of a plunger pump opening in a target reservoir stratum are collected; and inputting the monitoring data into the trained storage potential prediction model, and evaluating the storage potential of the target reservoir based on the long-term pressure change trend of the storage potential prediction model. According to the invention, the storage potential of the reservoir can be effectively estimated.

Description

Technical Field

[0001] This application relates to the technical field of mine water treatment, and particularly to a method and system for predicting the storage potential based on the deep geological storage technology of mine water. Background Art

[0002] The off-site deep storage technology for high salinity mine water and high saline water in coal mines provides a practical technical solution for the coal mining industry to solve the "last mile" problem in mine water treatment. When carrying out the deep storage of mine water, the water storage potential of the reservoir has become a crucial indicator in geological storage technology. However, most of the existing evaluation methods mainly rely on numerical simulation technology. Although these methods have certain guiding significance in theory, there are often large deviations from the actual storage potential in practical applications. Summary of the Invention

[0003] The purpose of this application is to provide a method and system for predicting the storage potential based on the deep geological storage technology of mine water, which can effectively estimate the storage potential of the reservoir.

[0004] To achieve the above purpose, this application provides the following solutions: In the first aspect, this application provides a method for predicting the storage potential based on the deep geological storage technology of mine water, including: Collect the monitoring data at the plunger pump port of the target reservoir; the monitoring data includes pressure monitoring data and flow monitoring data.

[0005] Input the monitoring data into the trained storage potential prediction model, and evaluate the storage potential of the target reservoir based on the long-term pressure change trend of the storage potential prediction model.

[0006] Among them, the training method of the storage potential prediction model is: Collect the sample monitoring data, geological exploration data, and core analysis data at the plunger pump ports of several reservoirs.

[0007] Fill in the missing values and remove the outliers from the sample monitoring data to obtain the preprocessed sample monitoring data.

[0008] Classify the sample monitoring data into the breakthrough period and the injection period according to the dispersion degree of the preprocessed sample monitoring data.

[0009] Fuse the sample monitoring data, geological exploration data, and core analysis data in the breakthrough period and the injection period to generate a comprehensive data set.

[0010] Train the storage potential prediction model based on the comprehensive data set to obtain the trained storage potential prediction model.

[0011] Optionally, the geological exploration data parameters include rock porosity, rock permeability, formation pressure, and rock saturation; the core analysis test data parameters include rock mineral composition, rock mechanical properties, and rock microstructure.

[0012] Optionally, based on the comprehensive data set, the storage potential prediction model is trained to obtain a trained storage potential prediction model, which specifically includes: Based on the data set, parallel training is performed on the ARIMA model, XGBOOST model, and LSTM model.

[0013] Based on the geological adaptability evaluation index, the trained ARIMA model, XGBOOST model, and LSTM model are optimally screened to obtain a trained storage potential prediction model.

[0014] Optionally, the missing value filling adopts the moving average method with a dynamic window to dynamically adjust the data difference before and after the sample monitoring data.

[0015] Optionally, the geological adaptability evaluation index includes prediction errors under different porosities, permeabilities, formation pressures, rock saturations, mineral compositions, mechanical properties, and microstructural conditions.

[0016] Optionally, the parameters to be trained in the ARIMA model include p, d, q, where P is the order of the autoregressive part, d is the order of the integration part, and q is the order of the moving average part; the parameters to be trained in the XGBOOST model include n_estimators, earning_rate, max_depth, subsample, colsample_bytree, and l2_lambda; the parameters to be trained in the LSTM model include hidden_size, num_epochs, and batch_siz.

[0017] Optionally, based on the long-term pressure change trend of the storage potential prediction model, the storage potential of the target reservoir is evaluated, which specifically includes: Predict the long-term pressure change trend based on the storage potential prediction model.

[0018] According to the predicted long-term pressure change trend, calculate the injection cycle experienced until the rated pressure of the plunger pump, and evaluate the storage potential of the reservoir.

[0019] In a second aspect, the present application provides a storage potential prediction system based on the deep geological storage technology of mine water depth, including: A data acquisition module for collecting monitoring data at the plunger pump port of the target reservoir; the monitoring data includes pressure monitoring data and flow monitoring data.

[0020] A prediction module for inputting the monitoring data into a trained storage potential prediction model and evaluating the storage potential of a target reservoir based on the long-term pressure change trend of the storage potential prediction model.

[0021] The training module includes: A sample data unit for collecting sample monitoring data, geological exploration data, and core analysis data at the plunger pump ports of several reservoirs.

[0022] A preprocessing unit for filling in missing values and removing outliers from the sample monitoring data to obtain preprocessed sample monitoring data.

[0023] A classification unit for classifying the sample monitoring data into a breakthrough period and an injection period according to the dispersion degree of the preprocessed sample monitoring data.

[0024] A data fusion unit for fusing the sample monitoring data, geological exploration data, and core analysis data in the breakthrough period and the injection period to generate a comprehensive data set.

[0025] A training unit for training the storage potential prediction model based on the comprehensive data set to obtain a trained storage potential prediction model.

[0026] Optionally, the training unit includes: A parallel training subunit for parallelly training the ARIMA model, the XGBOOST model, and the LSTM model based on the data set.

[0027] A screening subunit for optimally screening the trained ARIMA model, the XGBOOST model, and the LSTM model based on geological adaptability evaluation indicators to obtain a trained storage potential prediction model.

[0028] Optionally, the prediction module includes: A change trend prediction unit for predicting the long-term pressure change trend based on the storage potential prediction model.

[0029] A storage potential calculation unit for calculating the injection cycle experienced until the rated pressure of the plunger pump according to the predicted long-term pressure change trend and evaluating the storage potential of the reservoir.

[0030] According to the specific embodiments provided by the present application, the present application discloses the following technical effects: The present application provides a method and system for predicting the storage potential based on the deep geological storage technology of mine water depth. First, by collecting the monitoring data at the plunger pump port of the target reservoir, including pressure monitoring data and flow monitoring data, these data can reflect the dynamic changes of the reservoir during the deep geological storage process of mine water depth. Subsequently, these monitoring data are input into a pre-trained storage potential prediction model. This model evaluates the storage potential of the target reservoir based on the long-term pressure change trend, which can more accurately reflect the pressure response characteristics of the reservoir during the long-term storage process. During the training process of the storage potential prediction model, sample monitoring data, geological exploration data, and core analysis data of several reservoirs are collected, and these data cover various geological and engineering characteristics of the reservoir. By preprocessing the sample monitoring data, such as filling missing values and removing outliers, the quality and reliability of the data are ensured. Further, according to the dispersion degree of the preprocessed sample monitoring data, the sample data are classified into the breakthrough period and the injection period, which helps to better understand the behavioral characteristics of the reservoir at different storage stages. The sample monitoring data, geological exploration data, and core analysis data in the breakthrough period and the injection period are fused to generate a comprehensive data set. Based on the comprehensive data set, the storage potential prediction model is trained, enabling the model to learn the complex relationship between the storage potential of the reservoir and various geological and engineering characteristics. Therefore, when new monitoring data are input, the trained storage potential prediction model can accurately evaluate the storage potential of the target reservoir. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0032] Figure 1 It is a schematic flow chart of a method for predicting the storage potential based on the deep geological storage technology of mine water depth provided by an embodiment of the present application.

[0033] Figure 2 It is a schematic training flow chart provided by an embodiment of the present application.

[0034] Figure 3 It is the prediction results under different cycles provided by an embodiment of the present application.

[0035] Figure 4 It is a schematic diagram of the functional modules of a system for predicting the storage potential based on the deep geological storage technology of mine water depth provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0037] Embodiment 1 As Figure 1 shown, this embodiment provides a prediction method for the storage potential based on the deep geological storage technology of mine water depth, including: Step 101: Collect the monitoring data at the plunger pump port of the target reservoir; the monitoring data includes pressure monitoring data and flow monitoring data.

[0038] Step 102: Input the monitoring data into the trained storage potential prediction model, and evaluate the storage potential of the target reservoir based on the long-term pressure change trend of the storage potential prediction model.

[0039] Among them, as Figure 2 shown, the training method of the storage potential prediction model is: Step 201: Collect the sample monitoring data, geological exploration data, and core analysis data at the plunger pump ports of several reservoirs.

[0040] Step 202: Fill in the missing values and remove the outliers from the sample monitoring data to obtain the preprocessed sample monitoring data.

[0041] Step 203: Classify the sample monitoring data into the breakthrough period and the injection period according to the dispersion degree of the preprocessed sample monitoring data.

[0042] Step 204: Integrate the sample monitoring data, geological exploration data, and core analysis data in the breakthrough period and the injection period to generate a comprehensive data set.

[0043] Step 205: Train the storage potential prediction model based on the comprehensive data set to obtain the trained storage potential prediction model.

[0044] Among them, in some embodiments, when performing steps 201-205, it can be specifically as follows: Data collection: Collect the original data automatically monitored at the plunger pump port, including parameters such as pressure and flow, and the recording time interval is 5 minutes.

[0045] Data preprocessing includes missing value processing, outlier processing, and data classification, and it can be specifically as follows: Missing value handling: The `isnull` function is used to identify and locate missing values in the data at each stage. For missing values, the moving average method is adopted for filling, and the size of the filling window is determined according to the difference between the front and back data, so as to fully reduce the sharp fluctuations of the data signal while retaining the long-term trend.

[0046] Outlier handling: Data exceeding 1.5 times the interquartile range of the upper quartile or below the lower quartile is regarded as an outlier, and rows containing missing values and outliers are all deleted.

[0047] Data classification: Calculate the median, arithmetic mean, geometric mean, and harmonic mean of each group, classify the dispersion degree of the data in each group according to the results, classify each group as a "breakthrough period" and an "injection period", and divide the "breakthrough period" into several stages according to monotonicity.

[0048] Then, construct a prediction model for the sealing potential, which can be specifically as follows: 1) Model selection: To achieve time series prediction with underground pressure as a single variable, a comprehensive strategy including traditional statistical methods, machine learning tree models, and deep learning methods is adopted, and the availability of several types of methods in predicting the pressure change during the prediction stage is compared. Specifically, select the following three models: ARIMA model: The parameters are selected as p = 5, d = 1, q = 0.

[0049] XGBOOST model: The parameters are selected as n_estimators = 1000, learning_rate = 0.01, max_depth = 6, subsample = 0.8, colsample_bytree = 0.8, l2_lambda = 0.1.

[0050] LSTM model: The parameters are selected as input_size = 1, hidden_size = 128, output_size = 1, l2_lambda = 0.001, optimizer = Adam, learning_rate = 0.001, num_epochs = 100, batch_size = 32.

[0051] In the model training stage, in addition to using the preprocessed automatic monitoring data at the plunger pump port, geological exploration data and core analysis and test data parameters obtained during logging need to be introduced to further improve the accuracy and reliability of the model.

[0052] Among them, the geological exploration data parameters include rock porosity, rock permeability, formation pressure, and rock saturation. The core analysis and test data parameters include rock mineral composition, rock mechanical properties, rock microstructure, etc.

[0053] 2) Data fusion and feature engineering: Fuse the above geological exploration data, core analysis and test data with the automatic monitoring data at the plunger pump inlet to form a comprehensive data set. Feature extraction and dimensionality reduction techniques can be used for data fusion to extract the features most valuable for predicting the storage potential. Perform feature engineering on the fused data, including steps such as feature normalization and feature selection. Feature normalization scales all feature values to the same range to eliminate the influence of different feature dimensions and value ranges on model training. Feature selection can be carried out through methods such as correlation analysis and principal component analysis (PCA) to screen out the features most relevant to the prediction of storage potential, reduce the complexity of the model, and improve the training efficiency.

[0054] 3) Model training and optimization: Use the fused data set to train the ARIMA, XGBOOST, and LSTM models respectively. During the training process, adjust the hyperparameters of each model according to its characteristics to obtain better prediction performance. For example, for the ARIMA model, according to the autocorrelation and partial autocorrelation plots of the data, further optimize the p, d, q parameters. The ARIMA model consists of three parts: autoregressive (AR), integration (I), and moving average (MA). The parameters p, d, q of the model represent the orders of these three parts respectively. P is the order of the autoregressive part, indicating the relationship between the current value and the previous p historical values. D is the order of the integration part, indicating the number of times the time series needs to be differenced to become a stationary series. Q is the order of the moving average part, indicating the relationship between the current value and the previous q historical error terms; for the XGBOOST model, adjust parameters such as n_estimators, learning_rate, max_depth through grid search and cross-validation; for the LSTM model, adjust parameters such as hidden_size, num_epochs, batch_size.

[0055] During the training process, after introducing the parameters of geological exploration data and core analysis and test data, the model needs to learn the complex relationship between these parameters and the storage potential of mine water. Through a large amount of training data and optimized training algorithms, the model can better understand and predict the storage capacity of the reservoir under different geological conditions.

[0056] Specifically, to further improve the generalization ability of the model, data augmentation techniques are adopted during the training process. By randomly perturbing the geological exploration data and interpolating the core analysis and test data, etc., multiple training samples are generated, enabling the model to better adapt to different geological conditions and data changes.

[0057] 4) Model performance evaluation: In the model performance evaluation stage, in addition to using indicators such as Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and Maximum Error (Max Error), evaluation indicators related to geological conditions can also be introduced to more comprehensively evaluate the prediction performance of the model.

[0058] Geological adaptability evaluation: Evaluate the prediction performance of the model under different geological conditions, including prediction errors under different porosities, permeabilities, formation pressures, rock saturations, mineral compositions, mechanical properties, and microstructural conditions. By comparing the prediction results under different geological conditions, analyze the sensitivity of the model to changes in geological parameters, and determine the applicability and reliability of the model in different geological scenarios.

[0059] 5) Comprehensive performance evaluation: Comprehensively consider factors such as the prediction accuracy, calculation duration, and geological adaptability of the model, and select the optimal model for sequestration potential prediction. For example, if a certain model is slightly lower than other models in prediction accuracy, but has significant advantages in calculation duration and geological adaptability, and its prediction error is within an acceptable range, then this model can also be regarded as the optimal model.

[0060] Finally, conduct sequestration potential prediction, which can be specifically as follows: Predict the long-term pressure change trend based on the sequestration potential prediction model.

[0061] According to the predicted long-term pressure change trend, calculate the injection cycles experienced until the rated pressure of the piston pump, and evaluate the sequestration potential of the reservoir.

[0062] Specifically, the prediction of the long-term pressure change trend is as follows: Use the model with the best performance to predict the long-term pressure change trend, with the prediction time range being [specific time range, such as the next 1 year], and the prediction step size being [specific step size, such as 1 day].

[0063] The calculation of the injection cycle is specifically as follows: According to the predicted pressure change trend, calculate the injection cycles experienced until the rated pressure of the piston pump, and evaluate the sequestration potential of the reservoir. The specific calculation method is: When the predicted pressure reaches the rated pressure of the piston pump, record the time point at this time. The time interval between this time point and the previous time point is an injection cycle. Count the number of injection cycles within the entire prediction time range to evaluate the sequestration potential of the reservoir.

[0064] Among them, in the specific implementation process, it is assumed that the automatic monitoring data of the plunger pump port of a certain mine, as well as the logging data and core analysis and test data of the mine have been collected. The logging data includes a reservoir porosity of 25%, a permeability of 50 mD, a formation pressure of 5 MPa, and a rock saturation of 60%; the core analysis and test data includes that the main rock mineral components are mudstone and sandstone, with a compressive strength of 50 MPa, a shear strength of 20 MPa, an elastic modulus of 5 GPa, and the pore size distribution in the range of 0.5 - 5 μm.

[0065] Fuse these geological exploration data and core analysis and test data with the automatic monitoring data of the plunger pump port to form a comprehensive data set. Then, according to the above-mentioned feature engineering and model training steps, train and optimize the ARIMA, XGBOOST, and LSTM models respectively.

[0066] For example, when training the XGBOOST model, divide the fused data set into a training set and a test set. The training set accounts for 80% of the total data, and the test set accounts for 20%. Through grid search and cross-validation, adjust the parameters of the XGBOOST model. Finally, the optimal parameters are determined as n_estimators = 1200, learning_rate = 0.015, max_depth = 7, subsample = 0.9, colsample_bytree = 0.9, l2_lambda = 0.15. The RMSE of the XGBOOST model trained with these parameters on the test set is 0.05 MPa, the MAPE is 2%, and the Max Error is 0.1 MPa, showing good prediction performance.

[0067] In terms of geological adaptability evaluation, by comparing the prediction results under different porosity, permeability, and other conditions, it is found that the model has a smaller prediction error and better geological adaptability under geological conditions with a porosity range of 20% - 30% and a permeability range of 30 - 70 mD.

[0068] Taking into account factors such as prediction accuracy, calculation duration, and geological adaptability, finally select the XGBOOST model as the optimal model for predicting the storage potential. Use the selected XGBOOST model to predict the long-term pressure change trend and calculate the injection cycle to evaluate the storage potential of the reservoir. The prediction results show that within the next 3 years, the reservoir of this mine can complete 3 injection cycles, providing a scientific prediction basis for the deep geological storage of mine water. As Figure 3 shown, are the prediction results for different cycles.

[0069] Example 2 As Figure 4As shown in the figure, this embodiment provides a prediction system for the storage potential based on the deep geological storage technology of mine water depth, including: A data acquisition module 401, configured to acquire the monitoring data at the plunger pump port of the target reservoir; the monitoring data includes pressure monitoring data and flow monitoring data.

[0070] A prediction module 402, configured to input the monitoring data into the trained storage potential prediction model, and evaluate the storage potential of the target reservoir based on the long-term pressure change trend of the storage potential prediction model.

[0071] The training module 403 includes: A sample data unit, configured to acquire the sample monitoring data, geological exploration data, and core analysis data at the plunger pump port of several reservoirs.

[0072] A preprocessing unit, configured to fill in the missing values and remove the outliers from the sample monitoring data, and obtain the preprocessed sample monitoring data.

[0073] A classification unit, configured to classify the sample monitoring data into a breakthrough period and an injection period according to the dispersion degree of the preprocessed sample monitoring data.

[0074] A data fusion unit, configured to fuse the sample monitoring data, geological exploration data, and core analysis data in the breakthrough period and the injection period to generate a comprehensive data set.

[0075] A training unit, configured to train the storage potential prediction model based on the comprehensive data set to obtain the trained storage potential prediction model.

[0076] Among them, the training unit includes: A parallel training subunit, configured to perform parallel training on the ARIMA model, the XGBOOST model, and the LSTM model based on the data set.

[0077] A screening subunit, configured to perform optimal screening on the trained ARIMA model, the XGBOOST model, and the LSTM model based on the geological adaptability evaluation index to obtain the trained storage potential prediction model.

[0078] Among them, the prediction module includes: A change trend prediction unit, configured to predict the long-term pressure change trend based on the storage potential prediction model.

[0079] A storage potential calculation unit, configured to calculate the injection cycle experienced until the rated pressure of the plunger pump according to the predicted long-term pressure change trend, and evaluate the storage potential of the reservoir.

[0080] In summary, the present application has the following technical effects: This application proposes a method for predicting the pressure change trend based on the automatic monitoring data of the plunger pump port, which can effectively evaluate the storage potential of the reservoir and provide technical support for deep geological storage.

[0081] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0082] Specific examples are used in this article to elaborate on the principle and implementation manner of this application. The description of the above embodiments is only used to help understand the method and its core idea of this application; at the same time, for those of ordinary skill in the art, according to the idea of this application, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to this application.

Claims

1. A method for predicting the storage potential based on the deep geological storage technology of mine water depth, characterized in that Including: Collecting the monitoring data at the plunger pump port of the target reservoir; the monitoring data includes pressure monitoring data and flow monitoring data; Inputting the monitoring data into the trained sequestration potential prediction model, and evaluating the sequestration potential of the target reservoir based on the long-term pressure change trend of the sequestration potential prediction model; Among them, the training method of the sequestration potential prediction model is: Collecting the sample monitoring data, geological exploration data and core analysis data at the plunger pump ports of several reservoirs; Performing missing value filling and outlier removal on the sample monitoring data to obtain the preprocessed sample monitoring data; Classifying the sample monitoring data into a breakthrough period and an injection period according to the dispersion degree of the preprocessed sample monitoring data; Fusing the sample monitoring data, geological exploration data and core analysis data in the breakthrough period and the injection period to generate a comprehensive data set; Training the sequestration potential prediction model based on the comprehensive data set to obtain the trained sequestration potential prediction model.

2. A prediction method for storage potential based on deep geological storage technology of mine water depth according to claim 1, characterized in that, The geological exploration data parameters include rock porosity, rock permeability, formation pressure and rock saturation; the core analysis test data parameters include rock mineral composition, rock mechanical properties and rock microstructure.

3. A method for predicting the storage potential based on the deep geological storage technology of mine water depth according to claim 2, characterized in that, Training the sequestration potential prediction model based on the comprehensive data set to obtain the trained sequestration potential prediction model, specifically including: Parallel training of the ARIMA model, XGBOOST model and LSTM model based on the data set; Based on the geological adaptability evaluation index, optimally screening the trained ARIMA model, XGBOOST model and LSTM model to obtain the trained sequestration potential prediction model.

4. A method for predicting the storage potential based on the deep geological storage technology of mine water depth according to claim 3, characterized in that, The missing value filling adopts the moving average method with a dynamic window to dynamically adjust the difference between the front and rear data of the sample monitoring data.

5. A method for predicting the storage potential based on the deep geological storage technology of mine water depth according to claim 4, characterized in that The geological adaptability evaluation index includes the prediction errors under different porosity, permeability, formation pressure, rock saturation, mineral composition, mechanical properties and microstructure conditions.

6. The prediction method for the storage potential based on the deep geological storage technology of mine water depth according to claim 5, wherein, The parameters to be trained in the ARIMA model include p, d, q, where P is the order of the autoregressive part, d is the order of the integration part, and q is the order of the moving average part; the parameters to be trained in the XGBOOST model include n_estimators, earning_rate, max_depth, subsample, colsample_bytree and l2_lambda; the parameters to be trained in the LSTM model include hidden_size, num_epochs and batch_siz.

7. A prediction method for storage potential based on the deep geological storage technology of mine water depth according to claim 6, characterized in that, Evaluating the sequestration potential of the target reservoir based on the long-term pressure change trend of the sequestration potential prediction model, specifically including: Predicting the long-term pressure change trend based on the sequestration potential prediction model; Calculating the injection cycle experienced until the rated pressure of the plunger pump according to the predicted long-term pressure change trend, and evaluating the sequestration potential of the reservoir.

8. A prediction system for storage potential based on deep geological storage technology of mine water depth, characterized in that, Including: A data acquisition module for collecting the monitoring data at the plunger pump port of the target reservoir; the monitoring data includes pressure monitoring data and flow monitoring data; A prediction module, configured to input the monitoring data into a trained storage potential prediction model, and evaluate the storage potential of a target reservoir based on the long-term pressure change trend of the storage potential prediction model; The training module includes: A sample data unit, configured to collect sample monitoring data, geological exploration data, and core analysis data at the plunger pump ports of several reservoirs; A preprocessing unit, configured to fill in missing values and remove outliers from the sample monitoring data to obtain preprocessed sample monitoring data; A classification unit, configured to classify the sample monitoring data into a breakthrough period and an injection period according to the dispersion degree of the preprocessed sample monitoring data; A data fusion unit, configured to fuse the sample monitoring data, geological exploration data, and core analysis data in the breakthrough period and the injection period to generate a comprehensive data set; A training unit, configured to train a storage potential prediction model based on the comprehensive data set to obtain a trained storage potential prediction model.

9. A prediction system for storage potential based on the deep geological storage technology of mine water depth according to claim 8, characterized in that, The training unit includes: A parallel training subunit, configured to perform parallel training on an ARIMA model, an XGBOOST model, and an LSTM model based on the data set; A screening subunit, configured to perform optimal screening on the trained ARIMA model, XGBOOST model, and LSTM model based on geological adaptability evaluation indicators to obtain a trained storage potential prediction model.

10. A prediction system for storage potential based on deep geological storage technology of mine water depth according to claim 9, characterized in that, The prediction module includes: A change trend prediction unit, configured to predict the long-term pressure change trend based on the storage potential prediction model; A storage potential calculation unit, configured to calculate the injection cycle experienced until the rated pressure of the plunger pump according to the predicted long-term pressure change trend, and evaluate the storage potential of the reservoir.

Images