Deep brine target region optimization method and system

Abstract

The invention provides a deep brine target region optimization method and system. The deep brine target region optimization method comprises the following steps: S1, data acquisition and pretreatment: determining data acquisition targets: brine distribution range and depth, brine chemical components and quality, brine resource reserves and recoverability; geological survey: collecting geological information through field investigation, drilling an underground rock core sample, and drawing a regional geological map; and earth data acquisition: seismic data, gravity data, magnetic data, terrestrial heat data, hydrological data and well drilling data. The method comprehensively integrates multi-source geological information, scientifically and quantitatively evaluates the mineralization potential of the deep brine, significantly improves the accuracy and efficiency of deep brine exploration, lays a solid foundation for efficient development and utilization of deep brine resources, and has wide application prospects. And a neural network model is utilized to capture a complex nonlinear relationship, and through cross validation and hyper-parameter optimization, the prediction precision (such as a target area hit rate and a resource quantity error) is remarkably higher than that of a traditional method.

Description

Technical Field

[0001] The present invention relates to the field of geological exploration technology, and in particular to a method and system for optimizing a deep brine target area. Background Art

[0002] Target area optimization usually refers to the selection of the most promising target areas from the discovered mineralized points, based on the comprehensive geological, geophysical and geochemical characteristics. Deep potassium brine is a recent discovery, and its mineralization prediction and target area optimization are under discussion. The screening and delineation of mineralization target areas are not only the new content of mineralization prediction work, but also the basic work that must be carried out in the early stage of general prospecting and the basis for decision-making. On the basis of prediction mark optimization, the target area is minimized and the research on the mineralization rate of the target area is strengthened to improve the ore-finding rate of drilling verification, improve the hit rate of discovering ore deposits (bodies), and improve the geological prospecting effect. Therefore, the delineation of the target area is based on the minimum area and the maximum mineralization rate as the basic criteria. Target area screening is to further judge the possibility of discovering ore deposits in each target area and rank the mineralization favorableness when the target area has been fixed. It is a process of selecting the best and discarding the worst.

[0003] However, deep brines are buried at great depths and face complex geological conditions. Traditional exploration methods are often inefficient and inaccurate, resulting in high exploration costs and low success rates. Previous mineral exploration efforts relied heavily on regional geological surveys, simple geophysical surveys, and a small number of drilling verifications. This made it difficult to systematically and comprehensively assess the mineralization potential of deep brines, effectively preventing the identification of the most valuable target areas, and resulting in a significant waste of manpower, material resources, and time. Summary of the Invention

[0004] (1) Technical problems solved

[0005] In response to the shortcomings of the existing technology, the present invention provides a deep brine target area optimization method and system, which solves the problem that traditional exploration methods are often inefficient and inaccurate, resulting in high exploration costs and low success rates.

[0006] (2) Technical solution

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for selecting a deep brine target area comprises the following steps:

[0008] S1. Data collection and preprocessing:

[0009] (1) Determine the data collection objectives: brine distribution range and depth, brine chemical composition and quality, brine resource reserves and recoverability;

[0010] (2) Geological survey: field surveys to collect geological information, drill underground core samples, and draw regional geological maps;

[0011] (3) Earth data acquisition: seismic data, gravity data, magnetic data, geothermal data, hydrological data, and drilling data;

[0012] (4) Data cleaning: noise removal, missing value filling, normalization, detrending, spatial alignment, and temporal alignment;

[0013] (5) Feature extraction: geological features, physical features, chemical features, geothermal features;

[0014] (6) Data statistics: Pearson correlation coefficient is used to count and save data;

[0015] S2, geological multi-factor analysis;

[0016] (1) Based on the collected geological data, conduct lithologic analysis to determine the distribution of sandstones and carbonate rocks with pore or fracture development conditions in the target area;

[0017] (2) Structural analysis focuses on fault and fold structures. Faults are brine migration channels. Their strike, dip, fault throw, and openness are analyzed to identify water-conducting faults that connect deep heat sources with shallow strata. At the same time, the morphology, axis, and wing characteristics of the folds are studied. The axis is often conducive to brine accumulation due to rock fragmentation and stress concentration. The interlayer cracks in the wing can also provide storage space for brine. The control effect of structure on brine mineralization is comprehensively evaluated.

[0018] S3, model construction and training:

[0019] (1) Select a fully connected neural network learning algorithm to construct a deep brine target prediction model;

[0020] (2) Use historical exploration data to train the model and optimize model parameters through cross-validation to improve the generalization ability of the model;

[0021] S4. Target prediction and analysis:

[0022] (1) Multi-source data of the target area are input into the trained model, and the model outputs the prediction results of the brine target area;

[0023] (2) Visualize the prediction results, generate a brine target area distribution map, and use three-dimensional modeling tools to display the underground structure and brine distribution;

[0024] S5, target area delineation and verification;

[0025] (1) Establish a comprehensive evaluation indicator system;

[0026] (2) Gridded value calculation and analysis;

[0027] (3) Geophysical verification;

[0028] (4) Drilling verification.

[0029] Preferably, said S1 also includes organizing hydrogeological data, covering regional groundwater levels, flow directions, and hydrochemical type distribution information, understanding regional hydrodynamic conditions, and inferring brine migration, convergence paths, and possible enrichment areas.

[0030] Preferably, in S2, rock physical property data is obtained by core analysis and well logging interpretation to screen out favorable lithologic sections with high porosity, high permeability and a certain thickness.

[0031] Preferably, the three-dimensional modeling tool used in S4 is mapGIS three-dimensional geological modeling tool.

[0032] A deep brine target area optimization system includes a data acquisition and preprocessing module, a geological multi-factor analysis module, a model construction and training module, a target area prediction and analysis module, and a target area delineation and verification module.

[0033] (3) Beneficial effects

[0034] The present invention provides a method and system for selecting a deep brine target area. It has the following beneficial effects:

[0035] 1. The present invention comprehensively integrates multi-source geological information, scientifically and quantitatively evaluates the mineralization potential of deep brine, significantly improves the accuracy and efficiency of deep brine exploration, and lays a solid foundation for the efficient development and utilization of deep brine resources. In addition, the neural network model is used to capture complex nonlinear relationships. Through cross-validation and hyperparameter optimization, the prediction accuracy (such as target area hit rate and resource error) is significantly higher than that of traditional methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is an overall flow chart of a deep brine target area optimization method and system proposed by the present invention;

[0037] Figure 2 This is a model construction and training flow chart for a deep brine target area optimization method and system proposed in the present invention. DETAILED DESCRIPTION

[0038] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0039] Example:

[0040] like Figure 1-2As shown, an embodiment of the present invention provides a method for selecting a deep brine target area, comprising the following steps:

[0041] S1. Data collection and preprocessing:

[0042] (1) Determine the data collection objectives: brine distribution range and depth, brine chemical composition and quality, brine resource reserves and recoverability;

[0043] (2) Geological survey: field surveys to collect geological information, drill underground core samples, and draw regional geological maps;

[0044] (3) Earth data acquisition: seismic data, gravity data, magnetic data, geothermal data, hydrological data, and drilling data;

[0045] Through literature research and regional geological maps, we can understand the basic information of the target area, such as the strata, structure, and lithology, and clarify the types of data that need to be collected (such as seismic, drilling, geochemistry, etc.) and their priorities. In order to improve the efficiency of geological data collection, we can use multispectral cameras or LiDAR to quickly obtain surface topography and vegetation cover information. At the same time, we can use synthetic aperture radar (SAR) or hyperspectral remote sensing to monitor surface deformation and mineral distribution. We use relational databases (such as PostgreSQL) to store structured data (such as drilling logs), and upload large-capacity data such as seismic and remote sensing to cloud platforms (such as AWSS3).

[0046] (4) Data cleaning: noise removal, missing value filling, normalization, detrending, spatial alignment, and temporal alignment;

[0047] (5) Feature extraction: geological features, physical features, chemical features, geothermal features;

[0048] (6) Data statistics: Pearson correlation coefficient is used to count and save data;

[0049] Store the above data in a relational database (MySQL, PostgreSQL), and use XML or JSON format to store metadata.

[0050] S2, geological multi-factor analysis;

[0051] (1) Based on the collected geological data, conduct lithologic analysis to determine the distribution of sandstones and carbonate rocks with pore or fracture development conditions in the target area;

[0052] (2) Structural analysis focuses on fault and fold structures. Faults are brine migration channels. Their strike, dip, fault throw, and openness are analyzed to identify water-conducting faults that connect deep heat sources with shallow strata. At the same time, the morphology, axis, and wing characteristics of the folds are studied. The axis is often conducive to brine accumulation due to rock fragmentation and stress concentration. The interlayer cracks in the wing can also provide storage space for brine. The control effect of structure on brine mineralization is comprehensively evaluated.

[0053] S3, model construction and training:

[0054] (1) Select a fully connected neural network learning algorithm to construct a deep brine target prediction model;

[0055] Determine the network structure: First, determine the number of layers in the neural network, the number of neurons in each layer, and the activation function. For example, a simple fully connected neural network may include an input layer, several hidden layers, and an output layer. The number of neurons in the input layer depends on the number of features in the input data, while the number of neurons in the output layer is usually related to the number of predicted targets. Common activation functions include ReLU, Sigmoid, and Tanh.

[0056] Initialize weights and biases: Initialize the weight matrix and bias vector for each layer of the neural network. You can use random initialization methods such as normal distribution initialization and uniform distribution initialization, or you can use some predefined initialization strategies such as Xavier initialization and Kaiming initialization to help the model converge faster.

[0057] Model training: Prepare the dataset: Collect and organize the datasets for training, validation, and testing. The dataset should contain input features and corresponding labels (in supervised learning). Data preprocessing, such as normalization and standardization, is usually required to improve the training effect and convergence speed of the model; Define the loss function: Select an appropriate loss function based on the task type. For example, for regression tasks, commonly used loss functions include mean squared error (MSE) and mean absolute error (MAE); for classification tasks, the cross entropy loss function is commonly used; Select the optimizer: The optimizer is used to update the weights and biases of the neural network to minimize the loss function. Common optimizers include stochastic gradient descent (SGD), Adagrad, Adadelta, RMSProp, Adam, etc.; Train the model: Train in an iterative manner. In each iteration, a batch of data is input into the neural network, forward propagation is performed to calculate the output result, and then the loss value is calculated according to the loss function. Next, the gradient is calculated using the backpropagation algorithm, and the optimizer is used to update the weights and biases based on the gradient. This process is repeated until the loss function converges or the preset number of training rounds is reached. Validate and adjust the model: During the training process, the validation set is regularly used to evaluate the performance of the model. Based on the results on the validation set, the model's hyperparameters, such as the learning rate, number of layers, and number of neurons, can be adjusted to avoid overfitting or underfitting and improve the model's generalization ability. Evaluate the model: Use the test set to perform a final evaluation of the trained model, and calculate evaluation indicators such as the model's accuracy, recall rate, and F1 value on the test set to measure the model's performance.

[0058] (2) Use historical exploration data to train the model and optimize the model parameters through cross-validation to improve the generalization ability of the model. The specific method is as follows: randomly divide the data set into K subsets of similar size, select one of the subsets as the validation set each time, and the remaining K-1 subsets as the training set. Repeat this process K times so that each subset has the opportunity to be used as the validation set. Finally, the K validation results are averaged to obtain the model evaluation index.

[0059] S4. Target prediction and analysis:

[0060] (1) Multi-source data of the target area are input into the trained model, and the model outputs the prediction results of the brine target area;

[0061] (2) Visualize the prediction results, generate a brine target area distribution map, and use three-dimensional modeling tools to display the underground structure and brine distribution;

[0062] S5, target area delineation and verification;

[0063] (1) Establish a comprehensive evaluation indicator system;

[0064] (2) Gridded value calculation and analysis;

[0065] (3) Geophysical verification;

[0066] (4) Drilling verification.

[0067] The target area is divided into high probability target area, medium probability target area and low probability target area:

[0068] High-probability target areas: rich brine resources are predicted and exploration is prioritized;

[0069] Medium probability target area: brine resources are predicted to be moderate, with low exploration priority;

[0070] Low-probability target area: brine resources are predicted to be small, so exploration is temporarily suspended;

[0071] Design drilling well locations in high-probability target areas and determine drilling depth and direction.

[0072] Drilling implementation: Conduct drilling operations to obtain underground rock core and brine samples, analyze the stratigraphic structure, porosity, and permeability of the rock core, analyze the chemical composition of the brine samples (such as salinity, ion concentration, pH value, etc.), compare the drilling results with the predicted results, evaluate the prediction accuracy, and adjust the model parameters or retrain the model based on the comparison results to improve the prediction accuracy.

[0073] S1 also includes the compilation of hydrogeological data, covering regional groundwater levels, flow directions, and hydrochemical type distribution information, understanding regional hydrodynamic conditions, and inferring brine migration, convergence paths, and possible enrichment areas.

[0074] In S2, rock physical property data is obtained through core analysis and well logging interpretation, and favorable lithologic sections with high porosity, high permeability and a certain thickness are screened out.

[0075] The three-dimensional modeling tool used in S4 is mapGIS three-dimensional geological modeling tool.

[0076] A deep brine target area optimization system includes a data acquisition and preprocessing module, a geological multi-factor analysis module, a model construction and training module, a target area prediction and analysis module, and a target area delineation and verification module.

[0077] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing deep brine target areas, characterized in that: The following steps are involved: S1. Data collection and preprocessing: (1) Determine the data collection objectives: brine distribution range and depth, brine chemical composition and quality, brine resource reserves and recoverability; (2) Geological survey: field surveys to collect geological information, drill underground core samples, and draw regional geological maps; (3) Earth data acquisition: seismic data, gravity data, magnetic data, geothermal data, hydrological data, and drilling data; (4) Data cleaning; Noise removal, missing value filling, normalization, detrending, spatial alignment, and temporal alignment; (5) Feature extraction; geological characteristics, physical characteristics, chemical characteristics, and geothermal characteristics; (6) Data statistics: Pearson correlation coefficient is used to count and save data; S2, geological multi-factor analysis; (1) Based on the collected geological data, conduct lithologic analysis to determine the distribution of sandstones and carbonate rocks with pore or fracture development conditions in the target area; (2) Structural analysis focuses on fault and fold structures. Faults are brine migration channels. Their strike, dip, fault throw, and openness are analyzed to identify water-conducting faults that connect deep heat sources with shallow strata. At the same time, the morphology, axis, and wing characteristics of folds are studied. The axis is often conducive to brine accumulation due to rock fragmentation and stress concentration. The interlayer cracks in the wing can also provide storage space for brine. The control effect of structure on brine mineralization is comprehensively evaluated. S3, model construction and training: (1) Select a fully connected neural network learning algorithm to construct a deep brine target prediction model; (2) Use historical exploration data to train the model and optimize model parameters through cross-validation to improve the generalization ability of the model; S4. Target prediction and analysis: (1) Multi-source data of the target area are input into the trained model, and the model outputs the prediction results of the brine target area; (2) Visualize the prediction results, generate a brine target area distribution map, and use three-dimensional modeling tools to display the underground structure and brine distribution; S5, target area delineation and verification; (1) Establish a comprehensive evaluation indicator system; (2) Gridded value calculation and analysis; (3) Geophysical verification; (4) Drilling verification.

2. A deep brine target area selection method and system according to claim 1, characterized in that: Said S1 also includes organizing hydrogeological data, covering regional groundwater levels, flow directions, and hydrochemical type distribution information, understanding regional hydrodynamic conditions, and inferring brine migration, convergence paths, and possible enrichment areas.

3. The method and system for selecting a deep brine target area according to claim 1, characterized in that: In the S2, rock physical property data is obtained by core analysis and well logging interpretation, and favorable lithologic sections with high porosity, high permeability and a certain thickness are screened out.

4. A deep brine target area selection method and system according to claim 1, characterized in that: The three-dimensional modeling tool used in S4 is mapGIS three-dimensional geological modeling tool.

5. The method and system for selecting a deep brine target area according to claim 1, characterized in that: In S5, the analytic hierarchy process is used to calculate and analyze various data.

6. A deep brine target area optimization system, characterized in that: It includes data acquisition and preprocessing module, geological multi-factor analysis module, model construction and training module, target area prediction and analysis module, and target area delineation and verification module.

Images