Enhanced hot dry rock geothermal system fluid heat transfer process prediction method based on ANN network and agent model

Abstract

The invention relates to the technical field of geothermal research, and discloses an enhanced hot dry rock geothermal system fluid heat transfer process prediction method based on an ANN network, and the method comprises the following steps: constructing an enhanced hot dry rock geothermal system fluid heat transfer process prediction proxy model; inputting fracture network parameters into the prediction model; and the prediction model outputs a temperature breakthrough curve to predict the heat transfer behavior in the complex geothermal system. According to the method, the problem of efficiency bottleneck in numerical simulation is solved, an intelligent, economical and efficient analysis tool is provided for development of the enhanced geothermal system, and the innovative method is beneficial for accelerating popularization and application of the geothermal energy development technology and promoting technical progress and industry development of renewable energy utilization.

Description

Technical Field

[0001] The present invention relates to the field of geothermal research technology, and in particular to a method and an agent model for predicting fluid heat transfer processes in an enhanced hot dry rock geothermal system based on an ANN network. Background Art

[0002] Enhanced Geothermal Systems (EGS) as a renewable energy utilization method has broad application prospects. In EGS development, hot dry rock (HDR) is the main carrier of geothermal resources. By injecting cold water to exchange heat energy with the rock mass, hot water is extracted to achieve energy development. However, the fluid flow and heat transfer process in the hot dry rock geothermal system are affected by complex geological conditions and flow field characteristics, and its dynamic characteristics need to be analyzed in detail through numerical simulation methods.

[0003] Currently, numerical simulation technology is the main tool for studying fluid flow and heat transfer processes in enhanced geothermal systems. By instantaneously solving the heat transfer and fluid flow control equations, the flow path of cold water in the fracture network, temperature changes, and its thermal impact on the surrounding rock mass can be obtained. However, such models usually require high-resolution spatial discretization and time step calculations to capture the transient characteristics of heat flow transfer and cold front advancement. This sophisticated calculation process is not only time-consuming, but also requires powerful computing resources.

[0004] In practical applications, in order to optimize well network layout, design injection and production strategies, and evaluate system thermal performance, it is often necessary to run the numerical model repeatedly. For example, in inversion analysis or optimization design, it may be necessary to run the simulation model thousands or even tens of thousands of times to cover different input parameter combinations. This calculation mode of repeatedly calling the numerical model significantly increases the computing time and resource consumption, becoming a bottleneck problem in EGS development.

[0005] In order to solve the above problems, surrogate model technology has gradually become an effective alternative to numerical simulation in recent years. By learning and fitting the input and output relationship of the numerical simulation model, the surrogate model can provide prediction results with sufficient accuracy while significantly reducing the computational cost. Artificial Neural Network (ANN), as the main method for constructing surrogate models, has been widely used in the modeling and optimization of complex systems due to its strong nonlinear fitting ability, good self-learning ability, and wide adaptability. However, in existing research, the development of ANN-based surrogate models for the flow and heat transfer process of the EGS hot dry rock system is still in its initial stage, and there is a lack of efficient surrogate models designed specifically for the multi-parameter and highly nonlinear characteristics of transient heat transfer. Summary of the invention

[0006] In order to solve the technical problems raised in the background technology, the present invention provides a method for predicting the fluid heat transfer process of an enhanced hot dry rock geothermal system based on an ANN network.

[0007] The present invention is implemented by the following technical solution: a method for predicting the fluid heat transfer process of an enhanced hot dry rock geothermal system based on an ANN network, comprising the following steps:

[0008] Step 1: Construct a predictive proxy model for fluid heat transfer process in an enhanced hot dry rock geothermal system;

[0009] Step 2, using the fracture network parameters as input to the prediction model;

[0010] Step 3: The prediction model outputs the temperature breakthrough curve to predict the heat transfer behavior in complex geothermal systems.

[0011] The construction of the prediction proxy model for fluid heat transfer process in the enhanced hot dry rock geothermal system includes the following steps:

[0012] By using a fracture network generation method, fracture network data are obtained, and the fracture network data parameters include the number of fractures, fracture length, fracture angle, and fracture spacing;

[0013] The flow and heat transfer process of cold water from the injection well through the fracture network to the production well is simulated by numerical model, the heat transfer behavior between the injection and production wells is recorded, and the temperature breakthrough curve data is obtained;

[0014] A training data set was constructed with fracture network geometric parameters as input data and temperature breakthrough curve data as output data. The data was divided into a sample set and a training set. The model was trained and finally a prediction model for fluid heat transfer process in strong hot dry rock geothermal system was obtained.

[0015] Optionally, the prediction model training step is as follows: loading fracture network parameters and BTCs data from the database constructed in step 3;

[0016] Data standardization: standardize input parameters and output BTCs data;

[0017] Divide the data set into training set and validation set;

[0018] Build a neural network model;

[0019] Training and validation; Training process: Batch training: The batch size is set to 2000, and the model parameters are updated in each batch; Training cycle: Set 400 epochs (the number of times the data set is fully traversed); Evaluate the validation set performance every 10 epochs of training, and monitor the learning progress of the model in real time.

[0020] Early stopping mechanism: Set an early stopping strategy to terminate training when the performance of the validation set no longer improves. Save the model parameters with the best performance on the validation set as a file for subsequent use;

[0021] Save the model; save the trained neural network model and its parameters as a binary file (such as .pth format); store the model structure and weight file in the file system for subsequent loading and application.

[0022] Optionally, the specific operation of data standardization is: using sklearn.preprocessing.StandardScaler to standardize the input parameters and output BTC data, subtracting the mean of each feature from the data and then dividing it by the standard deviation.

[0023] Optionally, the training set division is performed as follows: the first 12,000 data are selected from the database for model learning; validation set: 2,000 data immediately following the training set are selected for model performance evaluation.

[0024] Optionally, use the PyTorch deep learning framework to build a neural network model, the specific structure of which is as follows:

[0025] Input layer: The input is the fracture network parameters, and the specific dimension is consistent with the number of features of the fracture network;

[0026] Hidden layer: 8 hidden layers are set, and the number of nodes is 180, 160, 120, 100, 90, 60, 60 and 20 respectively;

[0027] The activation function uses LeakyReLU (the negative slope is set to 0.01) to enhance the nonlinear expression ability of the model;

[0028] Output layer: The output is the standardized BTC data, and the dimension is equal to the number of discrete points of the BTC data. The output layer usually does not use an activation function and directly regresses the BTC value;

[0029] Optimizer and loss function: Optimizer: Use Adam algorithm, and the learning rate is set to 1×10 -5 . Loss function: The mean square error (MSE) is used to minimize the square difference between the predicted value and the actual value.

[0030] Optionally, the fracture network data may be generated by using an existing fracture network data set, or by using a fracture network generation method, and the fracture network generation method includes the following steps:

[0031] Step 1.1: Parameter setting and initialization in MATLAB;

[0032] Step 1.2: Generate a fracture network;

[0033] Step 1.3, read the fracture network data generated in step 1.2, store the data in the three-dimensional matrix DataDFN_CNN, and save it in the format of DataDFN_CNN.mat;

[0034] Optionally, the crack network generation method comprises the following steps:

[0035] The operation of step 1.2 is as follows:

[0036] Step a, generating crack length;

[0037] Step b, generating crack angle;

[0038] Step c: random generation of crack positions;

[0039] Step d: shadow area detection and removal;

[0040] Step e, crack intersection requirements;

[0041] Step f, judging the connectivity of the fracture network;

[0042] Step g: Save the finally generated fracture network data in a matrix format of size Nfrac*4, where: Nfrac: represents the number of fractures generated, and each row represents a fracture; 4: represents the four key attributes of each fracture.

[0043] Optionally, the numerical model simulation is constructed by the following method;

[0044] Physical model establishment; fluid flow is based on Darcy's law, cubic law and hydraulic coupling between fractures and matrix to ensure dynamic equilibrium between fluids in different regions; heat transfer is based on convection-conduction coupling equation and heat exchange model between fractures and matrix;

[0045] Numerical discretization and solution methods; discretization methods include using the finite element method (FEM) to discretize the fluid and heat transfer equations, using a time-stepping scheme; the solution process solves the transient flow and heat transfer equations by configuring a stable iterative algorithm, such as a step-by-step method or a nonlinear solver, and optimizes the computational efficiency of large-scale numerical models through parallel computing and appropriate convergence criteria;

[0046] Set model parameters and boundary conditions;

[0047] A two-dimensional heat reservoir model is established to simulate the fracture network and matrix area in the geothermal system. The typical size of the domain is 200m×200m. The wellbore areas on both sides are set, which are injection wells and production wells respectively. The wellbore areas are given high permeability and high porosity.

[0048] Boundary conditions:

[0049] The left injection well was set with constant hydraulic head and low temperature (20°C) conditions to simulate the cold water injection process.

[0050] The right production well was set with constant hydraulic head and heat outflow conditions to simulate hot water collection.

[0051] The upper and lower boundaries are set as adiabatic boundaries to avoid heat exchange with the external environment.

[0052] The initial condition was set to a uniform temperature distribution of 100 °C for both the matrix and the fracture network.

[0053] Material Properties:

[0054] Fracture area: high conductivity and low porosity, emphasizing that fractures are the main channels for fluids.

[0055] Matrix area: low conductivity and high porosity, ensuring that thermal conduction is the main mode of heat transfer;

[0056] Thermophysical parameters such as thermal conductivity, specific heat capacity and density are set according to typical thermal reservoir lithology parameters.

[0057] Optionally, the output results of the numerical model simulation include: the dynamic distribution of the temperature field; the temperature breakthrough curve of each monitoring point; the path and expansion range of the cold water flow in the fracture network.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] The present invention can improve computing efficiency: the high computing cost of traditional numerical simulation seriously restricts the application of optimization design and inversion analysis, while the prediction model proposed in the present invention can complete the same prediction task with extremely low running time, significantly improving computing efficiency.

[0060] The prediction model proposed in the present invention supports large-scale optimization: the present invention can quickly complete thousands or even tens of thousands of simulation predictions, and system design parameter optimization and sensitivity analysis become more feasible, providing important support for the design and operation of enhanced geothermal systems.

[0061] The present invention can reduce resource consumption: the present invention significantly reduces the dependence on high-performance computing equipment, while saving operating energy and computing time, which helps to promote the economy and sustainability of geothermal development.

[0062] The present invention can expand application scenarios: the prediction model proposed in the present invention is suitable for real-time prediction, site simulation and rapid performance evaluation due to its high efficiency and ease of use, which makes dynamic decision-making in geothermal resource development possible.

[0063] The present invention not only solves the efficiency bottleneck problem in numerical simulation, but also provides an intelligent, cost-effective and efficient analysis tool for the development of enhanced geothermal systems. This innovative method helps to accelerate the promotion and application of geothermal energy development technology and promote technological progress and industry development in the use of renewable energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A sample file of the fracture network data proposed in the present invention;

[0065] Figure 2 A dynamic distribution diagram of the temperature field simulated by the present invention;

[0066] Figure 3 The temperature breakthrough curve of each monitoring point simulated by the present invention;

[0067] Figure 4 The effect diagram of using this model to de-normalize the prediction results and restore them to the real BTC value;

[0068] Figure 5 The flowchart of the prediction method proposed by the present invention. DETAILED DESCRIPTION

[0069] The present invention is further described below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that, under the premise of no conflict, the various embodiments or technical features described below can be arbitrarily combined to form a new embodiment.

[0070] Embodiment 1:

[0071] Reference Figure 1-5 , This scheme proposes a method for predicting the fluid heat transfer process of an enhanced hot dry rock geothermal system based on an ANN network, which includes the following steps:

[0072] Step 1: Construct a predictive proxy model for fluid heat transfer process in an enhanced hot dry rock geothermal system;

[0073] Step 2, using the fracture network parameters as input to the prediction model;

[0074] Step 3: The prediction model outputs the temperature breakthrough curve to predict the heat transfer behavior in complex geothermal systems.

[0075] The construction of the prediction proxy model for fluid heat transfer process in the enhanced hot dry rock geothermal system includes the following steps:

[0076] By using a fracture network generation method, fracture network data are obtained, and the fracture network data parameters include the number of fractures, fracture length, fracture angle, and fracture spacing;

[0077] The flow and heat transfer process of cold water from the injection well through the fracture network to the production well is simulated by numerical model, the heat transfer behavior between the injection and production wells is recorded, and the temperature breakthrough curve data is obtained;

[0078] Specifically, for each group of fracture networks, the temperature change curves of multiple monitoring points are recorded. Specifically, they include:

[0079] Time series: temperature distribution at different time points during the simulation;

[0080] Monitoring point setting: Temperature sensors are arranged at the outlet of the fracture network to monitor the thermal response during the cold water advancement process.

[0081] The output of the numerical model simulation is a number of temperature breakthrough curves. The key features of the curves (such as pre-cold breakthrough time and cooling rate) directly reflect the geometry and heat transfer characteristics of the fracture network.

[0082] A training data set was constructed with fracture network geometric parameters as input data and temperature breakthrough curve data as output data. The data was divided into a sample set and a training set. The model was trained and finally a prediction model for fluid heat transfer process in strong hot dry rock geothermal system was obtained.

[0083] Optionally, the prediction model training step is as follows: loading fracture network parameters and BTCs data from the database constructed in step 3;

[0084] Data standardization: standardize input parameters and output BTCs data;

[0085] Divide the data set into training set and validation set;

[0086] Build a neural network model;

[0087] Training and validation; Training process: Batch training: The batch size is set to 2000, and the model parameters are updated in each batch; Training cycle: Set 400 epochs (the number of times the data set is fully traversed); Evaluate the validation set performance every 10 epochs of training, and monitor the learning progress of the model in real time.

[0088] Early stopping mechanism: Set an early stopping strategy to terminate training when the performance of the validation set no longer improves. Save the model parameters with the best performance on the validation set as a file for subsequent use;

[0089] Save the model; save the trained neural network model and its parameters as a binary file (such as .pth format); store the model structure and weight file in the file system for subsequent loading and application.

[0090] Optionally, the specific operation of data standardization is: using sklearn.preprocessing.StandardScaler to standardize the input parameters and output BTC data, subtracting the mean of each feature from the data and then dividing it by the standard deviation.

[0091] Optionally, the training set division is performed as follows: the first 12,000 data are selected from the database for model learning; validation set: 2,000 data immediately following the training set are selected for model performance evaluation.

[0092] Optionally, use the PyTorch deep learning framework to build a neural network model, the specific structure of which is as follows:

[0093] Input layer: The input is the fracture network parameters, and the specific dimension is consistent with the number of features of the fracture network;

[0094] Hidden layer: 8 hidden layers are set, and the number of nodes is 180, 160, 120, 100, 90, 60, 60 and 20 respectively;

[0095] The activation function uses LeakyReLU (the negative slope is set to 0.01) to enhance the nonlinear expression ability of the model;

[0096] Output layer: The output is the standardized BTC data, and the dimension is equal to the number of discrete points of the BTC data. The output layer usually does not use an activation function and directly regresses the BTC value;

[0097] Optimizer and loss function: Optimizer: Use Adam algorithm, and the learning rate is set to 1×10 -5 . Loss function: The mean square error (MSE) is used to minimize the square difference between the predicted value and the actual value.

[0098] Optionally, the fracture network data may adopt an existing fracture network dataset.

[0099] Optionally, the numerical model simulation is constructed by the following method;

[0100] Physical model establishment; fluid flow is based on Darcy's law, cubic law and hydraulic coupling between fractures and matrix to ensure dynamic equilibrium between fluids in different regions; heat transfer is based on convection-conduction coupling equation and heat exchange model between fractures and matrix;

[0101] Numerical discretization and solution methods; discretization methods include using the finite element method (FEM) to discretize the fluid and heat transfer equations, using a time-stepping scheme; the solution process solves the transient flow and heat transfer equations by configuring a stable iterative algorithm, such as a step-by-step method or a nonlinear solver, and optimizes the computational efficiency of large-scale numerical models through parallel computing and appropriate convergence criteria;

[0102] Set model parameters and boundary conditions;

[0103] A two-dimensional heat reservoir model is established to simulate the fracture network and matrix area in the geothermal system. The typical size of the domain is 200m×200m. The wellbore areas on both sides are set, which are injection wells and production wells respectively. The wellbore areas are given high permeability and high porosity.

[0104] Boundary conditions:

[0105] The left injection well was set with constant hydraulic head and low temperature (20°C) conditions to simulate the cold water injection process.

[0106] The right production well was set with constant hydraulic head and heat outflow conditions to simulate hot water collection.

[0107] The upper and lower boundaries are set as adiabatic boundaries to avoid heat exchange with the external environment.

[0108] The initial condition was set to a uniform temperature distribution of 100 °C for both the matrix and the fracture network.

[0109] Material Properties:

[0110] Fracture area: high conductivity and low porosity, emphasizing that fractures are the main channels for fluids.

[0111] Matrix area: low conductivity and high porosity, ensuring that thermal conduction is the main mode of heat transfer;

[0112] Thermophysical parameters such as thermal conductivity, specific heat capacity and density are set according to typical thermal reservoir lithology parameters.

[0113] Optionally, the output results of the numerical model simulation include: dynamic distribution of temperature field (see example results Figure 2 ); Temperature breakthrough curves for each monitoring point (BTCs results are shown in Figure 3 ); the path and extension of cold water flow in the fracture network.

[0114] Embodiment 2: In this embodiment, referring to Figure 1 ,, can be generated by a crack network generation method, the crack network generation method comprising the following steps:

[0115] Step 1.1: Parameter setting and initialization in MATLAB;

[0116] Step 1.2: Generate a fracture network;

[0117] Step 1.3, read the fracture network data generated in step 1.2, store the data in the three-dimensional matrix DataDFN_CNN, and save it in the format of DataDFN_CNN.mat;

[0118] Optionally, the crack network generation method comprises the following steps:

[0119] The operation of step 1.2 is as follows:

[0120] Step a, generating crack length;

[0121] Step b, generating crack angle;

[0122] Step c: random generation of crack positions;

[0123] Step d: shadow area detection and removal;

[0124] Step e, crack intersection requirements;

[0125] Step f, judging the connectivity of the fracture network;

[0126] Step g: Save the finally generated fracture network data in a matrix format of size Nfrac*4, where: Nfrac: represents the number of fractures generated, and each row represents a fracture; 4: represents the four key attributes of each fracture.

[0127] Specifically, the parameter setting and initialization operations are as follows:

[0128] Number of fractures: The number of fractures in each fracture network is randomly generated. The range of the number of fractures can be set according to the requirements, for example, between [14,20].

[0129] Crack angle: Set the crack direction parameters angle1 = -10 and angle2 = 65, representing the two main crack directions. std = 2 is the standard deviation of the crack angle.

[0130] Spatial dimensions: Define the spatial dimensions of the fracture network, xL=200, yL=200, and set the minimum fracture length lmin=50 and the maximum length lbig=140.

[0131] The operation of generating a fracture network includes the following steps.

[0132] Crack length generation

[0133] The crack length is generated based on the statistical model of fractal geometry, and the specific steps are as follows:

[0134] Parameter setting: Set the range between the minimum value lminl and the maximum value lbig of the crack length. The length of each crack is randomly generated within this range.

[0135] Distribution generation: The fractional dimension D and the power a are used to control the distribution of the crack length. The probability distribution of the crack length is calculated using the following formula.

[0136] N fra (L) = c·L -a

[0137] Where L is the length of the crack, a is the power that controls the crack length distribution, and c is the normalization constant. The length generated in this way follows a power law distribution, which can simulate the distribution of cracks of different lengths in nature.

[0138] Random length generation: Generate the length of each crack through the cumulative distribution function (CDF). The crack length is obtained by random number sampling to ensure the diversity of the crack network.

[0139] Crack angle generation; the crack angles are generated based on a normal distribution, with random deviations from two main directions:

[0140] Main angle: Set two main directions, namely θ 1 = -10° and θ 2 =65°.

[0141] Random deviation: The actual angle of each fracture is generated using a normal distribution with standard deviation σ = 2°. Some fractures are randomly deviated from the main direction to simulate the direction variation in the fracture network.

[0142] Angle normalization: Ensure that the generated angle is between 0° and 180° to avoid angles that are too large or too small.

[0143] Random generation of crack locations;

[0144] Random center point generation: The center point of each crack is randomly generated in the set domain [0, xL]×[0, yL]. The position follows a uniform distribution.

[0145] Specific region generation: For the first few cracks, a specific y-direction distribution can be set so that their centers are located in a specific region (e.g., near the center of the domain) to simulate a more orderly crack distribution.

[0146] Shadow area detection and removal;

[0147] After generating the cracks, it is necessary to detect whether the newly generated cracks invade the shadow area of ​​the existing cracks:

[0148] Shadow zone generation: A shadow zone is generated for each rift, usually a certain width (w, 2 unit length) around the rift, which represents the influence range of the rift.

[0149] Shadow area detection: Use the geometric intersection algorithm to detect whether the newly generated cracks overlap with the existing crack shadow areas. If overlap occurs, it means that the new crack generation position is not appropriate and the center position or angle needs to be regenerated.

[0150] Fissure intersection requirements;

[0151] Connectivity of the fracture network is achieved by enforcing that new fractures intersect with existing fractures:

[0152] Forced intersection: Each newly generated fracture is required to intersect with at least one existing fracture. This process is achieved by detecting the geometric intersection point of two fractures.

[0153] Geometric intersection detection: Use the geometric intersection algorithm to detect whether two cracks intersect. If the intersection requirement is not met, the algorithm will regenerate the center or angle of the crack until the intersection condition is met.

[0154] Determination of the connectivity of the fracture network;

[0155] The resulting fracture network must be connected from the left side of the domain to the right side:

[0156] Left-right connectivity detection: After the fracture network is generated, the algorithm will check whether there is a connected path from the left side of the domain x=0x=0x=0 to the right side x=xLx=xLx=xL. Only when the fracture network meets this connectivity will it be considered a qualified fracture network.

[0157] Connectivity determination: If the fracture network does not meet the connectivity requirements, the generation process will restart until a connected fracture network is generated.

[0158] The generated fracture network data will eventually be saved in a matrix format of size Nfrac*4.

[0159] Nfrac: represents the number of fractures generated, each row represents a fracture. 4: represents the four key attributes of each fracture, specifically: x1: the x-coordinate of the starting point of the fracture. y1: the y-coordinate of the starting point of the fracture. x2: the x-coordinate of the ending point of the fracture. y2: the y-coordinate of the ending point of the fracture. A total of 9,000 such fracture network samples were generated as training data.

[0160] The following table shows an example. Figure 1 shown.

[0161]

[0162] The present invention significantly improves the numerical simulation efficiency of the hot dry rock heat transfer process of the enhanced geothermal system by developing an efficient proxy model based on an artificial neural network (ANN). The proxy model uses the powerful nonlinear fitting ability of ANN to learn and approximate the complex relationship between the input parameters and output results of the numerical simulation model, thereby significantly reducing the calculation cost while ensuring accuracy. Compared with the traditional numerical simulation method that requires complex transient solutions, the proxy model of the present invention has higher operating efficiency. Originally, a complete numerical simulation took 2 to 5 minutes (the specific time consumption depends on the model scale and calculation conditions), but after using the ANN proxy model, each prediction only takes about 0.001 to 0.01 seconds to complete. The extremely high operating efficiency of the proxy model makes large-scale parameter optimization, well network layout design and thermal efficiency analysis possible.

[0163] The above-mentioned embodiments are only preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and substitutions made by technicians in this field on the basis of the present invention shall fall within the scope of protection required by the present invention.

Claims

1. A method for predicting fluid heat transfer process in an enhanced hot dry rock geothermal system based on an ANN network, characterized in that: The steps include: Step 1: Construct a predictive proxy model for fluid heat transfer process in an enhanced hot dry rock geothermal system; Step 2, using the fracture network parameters as input to the prediction model; Step 3: The prediction model outputs the temperature breakthrough curve to predict the heat transfer behavior in complex geothermal systems.

2. The enhanced hot dry rock geothermal system fluid heat transfer process prediction agent model according to claim 1, characterized in that: Constructing the prediction model includes the following steps: By using a fracture network generation method, fracture network data are obtained, and the fracture network data parameters include the number of fractures, fracture length, fracture angle, and fracture spacing; The flow and heat transfer process of cold water from the injection well through the fracture network to the production well is simulated by numerical model, the heat transfer behavior between the injection and production wells is recorded, and the temperature breakthrough curve data is obtained; A training data set was constructed with fracture network geometric parameters as input data and temperature breakthrough curve data as output data. The data was divided into a sample set and a training set. The model was trained and finally a prediction model for fluid heat transfer process in strong hot dry rock geothermal system was obtained.

3. The enhanced hot dry rock geothermal system fluid heat transfer process prediction agent model according to claim 2, characterized in that: The steps of training the prediction model are as follows: loading fracture network parameters and BTCs data from the database; Data standardization: standardize input parameters and output BTCs data; Divide the data set into training set and validation set; Build a neural network model; Training and validation; Training process: Batch training: The batch size is set to 2000, and the model parameters are updated in each batch; Training cycle: 400 epochs are set; The validation set performance is evaluated every 10 epochs, and the learning progress of the model is monitored in real time. Early stopping mechanism: Set an early stopping strategy to terminate training when the performance of the validation set no longer improves; save the model parameters with the best performance on the validation set as a file for subsequent use; Save the model; save the trained neural network model and its parameters as a binary file; store the model structure and weight file in the file system for subsequent loading and application.

4. The enhanced hot dry rock geothermal system fluid heat transfer process prediction agent model according to claim 2, characterized in that: The specific operation of data standardization is: use sklearn.preprocessing.StandardScaler to standardize the input parameters and output BTC data, subtract the mean of each feature from the data and divide it by the standard deviation.

5. The enhanced hot dry rock geothermal system fluid heat transfer process prediction agent model according to claim 2, characterized in that: The training set division is performed as follows: the first 12,000 data are selected from the database for model learning; the validation set: 2,000 data immediately following the training set are selected for model performance evaluation.

6. The enhanced hot dry rock geothermal system fluid heat transfer process prediction agent model according to claim 1, characterized in that: Use the PyTorch deep learning framework to build a neural network model. Its specific structure is as follows: Input layer: The input is the fracture network parameters, and the specific dimension is consistent with the number of features of the fracture network; Hidden layer: 8 hidden layers are set, and the number of nodes is 180, 160, 120, 100, 90, 60, 60 and 20 respectively; The activation function uses LeakyReLU to enhance the nonlinear expression ability of the model; Output layer: The output is standardized BTC data, and the dimension is equal to the number of discrete points of BTC data. The output layer usually does not use an activation function and directly regresses the BTC value. Optimizer and loss function: Optimizer: Use Adam algorithm, and the learning rate is set to 1×10 -5 . Loss function: Use mean square error to minimize the square difference between the predicted value and the actual value.

7. The enhanced hot dry rock geothermal system fluid heat transfer process prediction agent model according to claim 1, characterized in that: The fracture network data may be generated by using an existing fracture network data set, or by using a fracture network generation method, and the fracture network generation method includes the following steps: Step 1.1: Parameter setting and initialization in MATLAB; Step 1.2: Generate a fracture network; Step 1.3, read the fracture network data generated in step 1.2, store the data into the three-dimensional matrix DataDFN_CNN, and save it in the format of DataDFN_CNN.mat.

8. The enhanced hot dry rock geothermal system fluid heat transfer process prediction agent model according to claim 1, characterized in that: The crack network generation method comprises the following steps: The operation of step 1.2 is as follows: Step a, generating crack length; Step b, generating crack angle; Step c: random generation of crack positions; Step d: shadow area detection and removal; Step e, crack intersection requirements; Step f, judging the connectivity of the fracture network; Step g: Save the finally generated fracture network data in a matrix format of size Nfrac*4, where: Nfrac: represents the number of fractures generated, and each row represents a fracture; 4: represents the four key attributes of each fracture.

9. The enhanced hot dry rock geothermal system fluid heat transfer process prediction agent model according to claim 1, characterized in that: The numerical model simulation is constructed by the following method; Physical model establishment; fluid flow is based on Darcy's law, cubic law and hydraulic coupling between fractures and matrix to ensure dynamic equilibrium between fluids in different regions; heat transfer is based on convection-conduction coupling equation and heat exchange model between fractures and matrix; Numerical discretization and solution methods; discretization methods include using finite element method to discretize fluid and heat transfer equations, using time stepping scheme; the solution process solves transient flow and heat transfer equations by configuring stable iterative algorithms, such as step-by-step iteration method or nonlinear solver, and optimizes the computational efficiency of large-scale numerical models through parallel computing and appropriate convergence criteria; Set model parameters and boundary conditions; A two-dimensional heat reservoir model is established to simulate the fracture network and matrix area in the geothermal system. The typical size of the domain is 200m×200m. The wellbore areas on both sides are set, which are injection wells and production wells respectively. The wellbore areas are given high permeability and high porosity. Boundary conditions: The injection well on the left was set with constant hydraulic head and low temperature conditions to simulate the cold water injection process. The right production well was set with constant hydraulic head and heat outflow conditions to simulate hot water collection. The upper and lower boundaries are set as adiabatic boundaries to avoid heat exchange with the external environment. The initial condition was set to a uniform temperature distribution of 100 °C for both the matrix and the fracture network. Material Properties: Fracture area: high conductivity and low porosity, emphasizing that fractures are the main channels for fluids. Matrix area: low conductivity and high porosity, ensuring that thermal conduction is the main mode of heat transfer; Thermophysical parameters such as thermal conductivity, specific heat capacity and density are set according to typical thermal reservoir lithology parameters.

10. The enhanced hot dry rock geothermal system fluid heat transfer process prediction agent model according to claim 1, characterized in that: The output results of the numerical model simulation include: the dynamic distribution of the temperature field; the temperature breakthrough curve of each monitoring point; the path and expansion range of cold water flow in the fracture network.

Images