Prediction method and device for effective convection volume of in-situ leaching uranium mining

Abstract

The invention discloses a method and a device for predicting the effective convection volume of in-situ leaching uranium mining, relates to the technical field of nuclear energy, and mainly aims to solve the problem of poor prediction precision of the effective convection volume of the existing in-situ leaching uranium mining. Comprising the steps that the well type, the well spacing, the permeability coefficient and the liquid extraction amount of in-situ leaching uranium mining are obtained; a volume prediction model corresponding to the well type is called, the volume prediction model is obtained through training based on a constructed time volume training sample set, and effective convection volume samples in the time volume training sample set are obtained through simulation based on injection of multiple tracking particles into a three-dimensional model; and performing prediction processing on the well spacing, the permeability coefficient and the liquid pumping amount based on the volume prediction model to obtain an effective convective volume prediction value.

Description

Technical Field

[0001] The present application relates to the field of nuclear energy technology, and in particular to a method and device for predicting the effective convection volume of in-situ leaching uranium. Background Art

[0002] In-situ uranium leaching is an in-situ mining technology that relies on drilling to inject chemical solutions into mineral-bearing strata, controlling the hydraulic gradient of the flow field to allow the solution to migrate along the mineral layer and react with the ore to form a uranium-containing solution, which is then pumped out to the surface through a pumping well for separation and purification. In the process of in-situ uranium leaching, different well networks (well types, well spacing), geological conditions (such as permeability, porosity) and production operating conditions (such as injection volume, pumping volume) will significantly affect the leaching effect.

[0003] At present, the existing prediction of the effective convection volume that characterizes the leaching effect mainly relies on experimental experience and physical models, that is, the flow field simulation of in situ leaching of uranium is carried out by means of numerical simulation, and finally data analysis and post-processing are carried out. However, it is difficult to predict the accurate leaching effect based on experimental experience and physical models, and it takes a long time. In addition, it is impossible to meet the high-precision requirements of in situ leaching of uranium under different mining environments, which greatly affects the actual output. Summary of the invention

[0004] In view of this, the present application provides a method and device for predicting the effective convection volume of in-situ uranium leaching, the main purpose of which is to solve the problem of poor prediction accuracy of the effective convection volume of in-situ uranium leaching.

[0005] According to one aspect of the present application, a method for predicting the effective convection volume of in-situ leaching uranium is provided, comprising:

[0006] Obtain the well type, well spacing, permeability coefficient and pumping volume for in-situ leaching of uranium;

[0007] Retrieving a volume prediction model corresponding to the well type, wherein the volume prediction model is obtained by training based on a constructed time volume training sample set, and the effective convection volume samples in the time volume training sample set are obtained by simulation based on injecting multiple tracking particles into a three-dimensional model;

[0008] The well spacing, the permeability coefficient and the pumping volume are predicted based on the volume prediction model to obtain a predicted value of the effective convection volume.

[0009] Furthermore, before obtaining the well type, well spacing, permeability coefficient and liquid extraction volume of in situ leaching uranium, the method further includes:

[0010] Obtain samples of well spacing, permeability coefficient, and pumping volume for different well types;

[0011] defining a three-dimensional model, and injecting multiple tracking particles into the three-dimensional model;

[0012] Numerical simulation is performed based on the multiple tracking particles and the well spacing samples, the permeability coefficient samples and the pumping volume samples to obtain an effective convection volume simulation value.

[0013] Furthermore, after performing numerical simulation based on the multiple tracking particles and the well spacing samples, the permeability coefficient samples and the pumping volume samples to obtain the effective convection volume simulation value, the method further includes:

[0014] Constructing a fitting curve based on the effective convection volume simulation value and the simulation time, wherein the fitting curve includes an amplitude coefficient, a tilt factor and an offset;

[0015] A time-volume training sample set is constructed based on the optimally solved amplitude coefficient, slope factor, offset, and well spacing samples, permeability coefficient samples, and pumping volume samples of different well types.

[0016] Furthermore, the well type is a five-point type, and before calling the volume prediction model corresponding to the well type, the method further includes:

[0017] Acquire first fitting parameters of the first well spacing sample, the first permeability coefficient sample, the first pumping volume sample and the effective convection volume sample of the five-point type from the time volume training sample set;

[0018] Constructing the first multi-layer feedforward neural network;

[0019] The first well spacing sample, the first permeability coefficient sample, and the first pumping volume sample are used as first model input parameters, the first fitting parameter is determined as a first model output parameter, and the first multi-layer feedforward neural network is trained to obtain the five-point first volume prediction model.

[0020] Furthermore, the well type is a seven-point type, and before calling the volume prediction model corresponding to the well type, the method further includes:

[0021] Acquire second fitting parameters of the second well spacing sample, the second permeability coefficient sample, the second pumping volume sample and the effective convection volume sample of the seven-point type from the time volume training sample set;

[0022] Construct a second multi-layer feedforward neural network;

[0023] The second well spacing sample, the second permeability coefficient sample, and the second pumping volume sample are used as second model input parameters, the second fitting parameter is determined as the second model output parameter, and the second multi-layer feedforward neural network is trained to obtain the seven-point second volume prediction model.

[0024] Furthermore, the method further comprises:

[0025] During the model training process, the deviation is calculated during the training process of the first multi-layer feedforward neural network or the second multi-layer feedforward neural network by using the mean square error as a loss function, and the model training is completed when the deviation is minimum;

[0026] The weight value of the first multi-layer feedforward neural network or the second multi-layer feedforward neural network is obtained by dynamically adjusting the sample loss value determined based on the callback function.

[0027] Furthermore, the method further comprises:

[0028] Based on the volume prediction model, the well type, well spacing, permeability coefficient and pumping volume are predicted and processed to obtain fitting parameters of effective convection volume;

[0029] The fitting parameters are integrated with the fitting curve to obtain the predicted values ​​of effective convection volume at different time points.

[0030] According to another aspect of the present application, a device for predicting the effective convection volume of in-situ leaching uranium is provided, comprising:

[0031] An acquisition module is used to obtain the well type, well spacing, permeability coefficient and pumping volume of in-situ uranium leaching;

[0032] A calling module, used for calling a volume prediction model corresponding to the well type, wherein the volume prediction model is obtained by training based on a constructed time volume training sample set, and the effective convection volume samples in the time volume training sample set are obtained by simulation based on injecting multiple tracking particles into a three-dimensional model;

[0033] A processing module is used to predict the well spacing, the permeability coefficient and the pumping volume based on the volume prediction model to obtain a predicted value of the effective convection volume.

[0034] Furthermore, the device further comprises: a definition module, a simulation module,

[0035] The acquisition module is also used to acquire well spacing samples, permeability coefficient samples and pumping volume samples of different well types;

[0036] The definition module defines a three-dimensional model and injects multiple tracking particles into the three-dimensional model;

[0037] The simulation module is used to perform numerical simulation based on the multiple tracking particles and the well spacing samples, the permeability coefficient samples and the pumping volume samples to obtain an effective convection volume simulation value.

[0038] Furthermore, the device also includes:

[0039] A first construction module is used to construct a fitting curve based on the effective convection volume simulation value and the simulation time, wherein the fitting curve includes an amplitude coefficient, a tilt factor and an offset;

[0040] The second construction module is used to construct a time volume training sample set based on the optimally solved amplitude coefficient, slope factor, offset, and well spacing samples, permeability coefficient samples, and pumping volume samples of different well types.

[0041] Furthermore, the well type is a five-point type, and the device further comprises: a fourth construction module, a first training module,

[0042] The acquisition module is further used to acquire first fitting parameters of the first well spacing sample, the first permeability coefficient sample, the first pumping volume sample and the effective convection volume sample of the five-point type from the time volume training sample set;

[0043] The third building module is used to build a first multi-layer feedforward neural network;

[0044] The first training module is used to use the first well spacing sample, the first permeability coefficient sample, and the first pumping volume sample as first model input parameters, determine the first fitting parameter as the first model output parameter, and perform model training on the first multi-layer feedforward neural network to obtain the five-point first volume prediction model.

[0045] Furthermore, the well type is a seven-point type, and the device further comprises: a fifth construction module, a second training module,

[0046] The acquisition module is further used to acquire second fitting parameters of the second well spacing sample, the second permeability coefficient sample, the second pumping volume sample and the effective convection volume sample of the seven-point type from the time volume training sample set;

[0047] The fourth building block is used to build a second multi-layer feedforward neural network;

[0048] The second training module is used to use the second well spacing sample, the second permeability coefficient sample, and the second pumping volume sample as second model input parameters, determine the second fitting parameter as the second model output parameter, and perform model training on the second multi-layer feedforward neural network to obtain the five-point second volume prediction model.

[0049] Furthermore, the training module is also used to calculate the deviation during the training of the first multi-layer feedforward neural network or the second multi-layer feedforward neural network by using the mean square error as a loss function during the model training process, and complete the model training when the deviation is minimum; wherein the weight value of the first multi-layer feedforward neural network or the second multi-layer feedforward neural network is obtained by dynamically adjusting the sample loss value determined based on the callback function.

[0050] Further,

[0051] The processing module is also used to predict the well type, well spacing, permeability coefficient and pumping volume based on the volume prediction model to obtain fitting parameters of the effective convection volume; integrate the fitting parameters with the fitting curve to obtain the predicted values ​​of the effective convection volume at different time points.

[0052] According to another aspect of the present application, a storage medium is provided, wherein at least one executable instruction is stored in the storage medium, and the executable instruction enables a processor to execute operations corresponding to the above-mentioned method for predicting the effective convection volume of in-situ leaching uranium.

[0053] According to another aspect of the present application, a terminal is provided, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other through the communication bus;

[0054] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the above-mentioned method for predicting the effective convection volume of uranium in situ leaching.

[0055] By means of the above technical solution, the technical solution provided by the embodiment of the present application has at least the following advantages:

[0056] The present application provides a method and device for predicting the effective convection volume of in-situ uranium leaching. Compared with the prior art, the embodiment of the present application obtains the well type, well spacing, permeability coefficient and pumping volume of in-situ uranium leaching; calls up a volume prediction model corresponding to the well type, wherein the volume prediction model is obtained by training based on a constructed time volume training sample set, and the effective convection volume samples in the time volume training sample set are obtained by simulation based on injecting multiple tracking particles into a three-dimensional model; based on the volume prediction model, the well spacing, the permeability coefficient and the pumping volume are predicted and processed to obtain a predicted value of the effective convection volume, thereby reducing the prediction time of the effective convection volume as a leaching effect, achieving a more accurate prediction of the leaching effect, and meeting the high-precision requirements for in-situ uranium leaching under different mining environments, thereby greatly improving the output in actual generation.

[0057] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0059] Figure 1 A flow chart of a method for predicting the effective convection volume of uranium in situ leaching provided in an embodiment of the present application is shown;

[0060] Figure 2 A schematic diagram showing a comparison between a predicted value of an effective convection volume and a true value provided in an embodiment of the present application is shown;

[0061] Figure 3 A block diagram of a device for predicting the effective convection volume of uranium in situ leaching provided in an embodiment of the present application is shown;

[0062] Figure 4 A schematic diagram of the structure of a terminal provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0063] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0064] The present application embodiment provides a method for predicting the effective convection volume of uranium in situ leaching, such as Figure 1 As shown, the method includes:

[0065] 101. Obtain the well type, well spacing, permeability coefficient and pumping volume for in situ leaching of uranium.

[0066] In the embodiment of the present application, the current execution end as the execution subject can be a local terminal device or a cloud server. The well types for in-situ uranium leaching include five-point type and seven-point type. The well spacing is the distance between the injection well and the extraction well in in-situ uranium leaching. The permeability coefficient is the permeability of in-situ uranium leaching as a geological condition, preferably the formation permeability coefficient. The extraction volume is the operation volume of extraction in the in-situ uranium leaching production, so as to mine the uranium ore layer (the lithology of the uranium ore layer includes mudstone, sandstone and coarse sandstone, etc.) in the in-situ uranium leaching mining area. The embodiment of the present application does not make specific restrictions. Among them, the well type, well spacing, permeability coefficient and extraction volume can be pre-configured based on the operation requirements when conducting in-situ uranium leaching, so as to be directly retrieved when predicting the effective convection volume.

[0067] 102. Retrieve a volume prediction model corresponding to the well type.

[0068] In the embodiment of the present application, the current execution end pre-trains the corresponding volume prediction model for different well types, and the volume prediction model is obtained by training based on the constructed time volume training sample set, wherein the effective convection volume sample in the time volume training sample set is obtained by simulation based on the injection of multiple tracking particles in the three-dimensional model, that is, through the actual mining area situation, the position of each injection well and each pumping well in the three-dimensional model is defined, and multiple tracking particles are injected into the three-dimensional model. The existing open source programs Modflow-2005 and Modpath7 software are used for numerical simulation to determine the position of each tracking particle at different times. Finally, according to the position of each tracking particle at different times, the position of the cell where each injection well is located, and the position of the cell where each pumping well is located, the effective convection volume (Ve) simulation value of the overall in situ uranium mining area at different time nodes is calculated as a sample of the time volume training sample set, which is not specifically limited in the embodiment of the present application.

[0069] It should be noted that when using samples in the time volume training sample set for model training, the well spacing, permeability coefficient, and pumping volume are used as model input parameters, and the output parameters can be the fitting parameters in the fitting curve based on the effective convection volume simulation value, so that the trained model can optimally learn the relationship between the effective convection volume and time.

[0070] 103. Based on the volume prediction model, the well spacing, the permeability coefficient and the pumping volume are predicted to obtain a predicted value of the effective convection volume.

[0071] In the embodiment of the present application, after the volume prediction model matching the well type is retrieved, the well spacing, permeability coefficient and pumping volume are used for prediction processing to obtain the effective convection volume prediction value, which is used as the operational basis for in situ leaching of uranium. The embodiment of the present application does not make specific limitations.

[0072] In another embodiment of the present application, for further definition and explanation, before the step of obtaining the well type, well spacing, permeability coefficient and pumping volume of in situ leaching uranium, the method further includes:

[0073] Obtain samples of well spacing, permeability coefficient, and pumping volume for different well types;

[0074] defining a three-dimensional model, and injecting multiple tracking particles into the three-dimensional model;

[0075] Numerical simulation is performed based on the multiple tracking particles and the well spacing samples, the permeability coefficient samples and the pumping volume samples to obtain an effective convection volume simulation value.

[0076] In order to realize the construction of an effective model prediction model to accurately simulate the various parameters of the in-situ leaching uranium mining area, the current execution end first obtains well spacing samples, permeability coefficient samples and pumping volume samples of different well types, and defines a three-dimensional model. Among them, the three-dimensional model can be a three-dimensional heterogeneous geological model established by the T-PROGS module in the GMS (Groundwater Modeling System) application. By rendering the three-dimensional model corresponding to the entire in-situ leaching uranium mining area, a three-dimensional heterogeneous geological model is obtained. The embodiment of this application is not specifically limited. Before defining the three-dimensional model, grid division is first performed. When dividing, the size of the grid can be consistent or inconsistent. The grid records the position of each injection well and the position of each pumping well. The embodiment of this application is not specifically limited. After the three-dimensional model is divided, multiple tracking particles can be injected into the three-dimensional model. By recording the position of the tracking particles in the three-dimensional model at different times, the position of each tracking particle at different times can be obtained. After obtaining the position of each tracking particle at different times, the effective convection field volume simulation value of the in-situ uranium mining area can be determined by combining the position of the cell where each injection well is located and the position of the cell where each extraction well is located. The embodiment of the present application does not make specific limitations.

[0077] In another embodiment of the present application, for further definition and explanation, after the step of performing numerical simulation based on the multiple tracking particles and the well spacing samples, the permeability coefficient samples and the pumping volume samples to obtain the effective convection volume simulation value, the method further includes:

[0078] Constructing a fitting curve based on the effective convection volume simulation value and the simulation time;

[0079] A time-volume training sample set is constructed based on the optimally solved amplitude coefficient, slope factor, offset, and well spacing samples, permeability coefficient samples, and pumping volume samples of different well types.

[0080] In order to construct a more effective training sample set and thus improve the learning effect of model training, the current execution end can construct a fitting curve between the effective convection volume simulation value and the simulation time based on a logarithmic function log3p1 function, that is, to perform curve fitting on the relationship between the effective convection volume simulation value and time under different parameter combinations. At this time, the fitting curve includes the amplitude coefficient, the tilt factor, and the offset. The specific fitting curve is expressed as:

[0081] Among them, a is the amplitude coefficient, which is used to determine the overall amplitude of the fitting curve and scale the output of the function. b is the tilt factor, which is used to adjust the influence of the linear increment of x on y. c is the offset, which is used to adjust the vertical position of the curve on the y-axis and represents the baseline or initial value of the data. t is the time node in the numerical simulation process.

[0082] It should be noted that after constructing the fitting curve, the amplitude coefficient, slope factor and offset are optimally solved based on the nonlinear least squares method, and the time volume training sample set is constructed based on the optimally solved amplitude coefficient, slope factor, offset and well spacing samples, permeability coefficient samples and pumping volume samples of different well types. In this process, the nonlinear least squares method is used to optimize the solution parameters a, b and c to maximize the regression coefficient of the training model and simulation data, and minimize the residual sum of squares, and finally obtain the Log3P1 fitting results under different situations.

[0083] In another embodiment of the present application, for further definition and explanation, before the step of calling the volume prediction model corresponding to the well type, the method further includes:

[0084] Acquire first fitting parameters of the first well spacing sample, the first permeability coefficient sample, the first pumping volume sample and the effective convection volume sample of the five-point type from the time volume training sample set;

[0085] Constructing the first multi-layer feedforward neural network;

[0086] The first well spacing sample, the first permeability coefficient sample, and the first pumping volume sample are used as first model input parameters, the first fitting parameter is determined as a first model output parameter, and the first multi-layer feedforward neural network is trained to obtain the five-point first volume prediction model.

[0087] In order to make the training of the volume prediction model targeted, thereby improving the accuracy of model training and scene adaptability, specifically, when the well type is a five-point type, specifically, obtain the first fitting parameters of the first well spacing sample, the first permeability coefficient sample, the first pumping volume sample, and the effective convection volume sample of the five-point type from the time volume training sample set. Among them, when obtaining the above data, it is necessary to delete the rows containing null values, convert the data frame into an array format, dataset_all is a two-dimensional array, the row is a five-point parameter combination, and the column is a parameter variable. At the same time, construct a first multi-layer feedforward neural network, and use the first well spacing sample, the first permeability coefficient sample, and the first pumping volume sample as the first model input parameter, and determine the first fitting parameter as the first model output parameter, so as to realize model training of the first multi-layer feedforward neural network and obtain the first volume prediction model of the five-point type. In a specific embodiment, the input data dataset_x includes five-point well spacing, permeability, and flow rate, and the output data dataset_y includes the a, b, and c values ​​of the effective convection volume and time fitting curve, so as to train the model and learn the optimal model parameter variables. At the same time, the MinMaxScaler class in the existing library sklearn can be used for normalization. Among them, the MinMaxScaler class maps the data to a specified range (usually [0,1]) by scaling to obtain better convergence and performance in the machine learning model. At the same time, the normalization parameters can be saved using persistence technology so that the results can be denormalized in the prediction stage to restore them to the true dimension.

[0088] It should be noted that when obtaining input data and output data from the sample set, the input data and output data can be divided into a training set (80% of the total data) and a validation set (20% of the total data). For multi-layer feedforward neural networks, a multi-layer feedforward neural network architecture can be adopted, and the input layer is set to 3 nodes, corresponding to the dimension of the input features. The number of neurons in the hidden layer is gradually reduced to 256, 128, 64, and 32, respectively, and the ReLU activation function is used to enhance the nonlinear expression ability of the multi-layer feedforward neural network model. The output layer contains 3 nodes, which directly map the target dimension corresponding to the predicted value. The weight initialization method can be the uniform method to ensure that the weight distribution of the initial state of the model is uniform.

[0089] In another embodiment of the present application, for further definition and explanation, before the step of calling the volume prediction model corresponding to the well type, the method further includes:

[0090] Acquire second fitting parameters of the second well spacing sample, the second permeability coefficient sample, the second pumping volume sample and the effective convection volume sample of the seven-point type from the time volume training sample set;

[0091] Construct a second multi-layer feedforward neural network;

[0092] The second well spacing sample, the second permeability coefficient sample, and the second pumping volume sample are used as second model input parameters, the second fitting parameter is determined as the second model output parameter, and the second multi-layer feedforward neural network is trained to obtain the seven-point second volume prediction model.

[0093] In order to make the training of the volume prediction model targeted, thereby improving the accuracy of model training and scene adaptability, specifically, when the well type is a seven-point type, specifically, obtain the second fitting parameters of the second well spacing sample, second permeability coefficient sample, second pumping volume sample and effective convection volume sample of the seven-point type from the time volume training sample set. Among them, when obtaining the above data, it is necessary to delete the rows containing null values ​​and convert the data frame into an array format. dataset_all is a two-dimensional array, with a row as a seven-point parameter combination and a column as a parameter variable. At the same time, a second multi-layer feedforward neural network is constructed, and the second well spacing sample, the second permeability coefficient sample, and the second pumping volume sample are used as the second model input parameters, and the second fitting parameters are determined as the second model output parameters, so as to realize model training of the second multi-layer feedforward neural network and obtain the second volume prediction model of the seven-point type. In a specific embodiment, the input data dataset_x includes the well spacing, permeability coefficient, and flow rate of the seven-point type, and the output data dataset_y includes the a, b, and c values ​​of the effective convection volume and time fitting curve, so as to train the model and learn the optimal model parameter variables. At the same time, the MinMaxScaler class in the existing library sklearn can be used for normalization. In the embodiment of the present application, the model training process for the five-point type and the seven-point type is the same, which will not be repeated here.

[0094] In another embodiment of the present application, for further definition and explanation, the steps further include:

[0095] During the model training process, the deviation is calculated during the training of the first multi-layer feedforward neural network or the second multi-layer feedforward neural network by using the indicator mean square error as the loss function, and the model training is completed when the deviation is minimum.

[0096] In order to improve the accuracy of model training and thus improve the effectiveness of effective convection volume prediction, when the current execution end is training the model, whether it is for the five-point type or the seven-point type, the deviation can be calculated during the training of the first multi-layer feedforward neural network or the second multi-layer feedforward neural network by using the mean square error as the loss function, and the model training is completed when the deviation is the smallest. Among them, the weight value of the first multi-layer feedforward neural network or the second multi-layer feedforward neural network is obtained by dynamically adjusting the sample loss value determined by the callback function. In the model training process of the specific implementation scenario, the number of samples in each batch (batch_size=4) and the number of training rounds (epochs=1800) can also be set, and cross-validation is performed in combination with the validation set to ensure that the model has good generalization ability. In addition, the callback function introduced in the model training is preferably a Checkpoint callback function, which dynamically saves the weight file with the best performance according to the validation set loss value as a basis for dynamic adjustment, thereby ensuring the optimality and stability of the training results. After the training is completed, the multi-layer feedforward neural network model performs prediction verification on the training set and validation set data, and restores the prediction results to the actual physical dimensions through denormalization, which facilitates direct analysis and evaluation of the output parameters.

[0097] It should be noted that after the model training is completed, the five-point and seven-point volume prediction models are stored. When the effective convection volume of the five-point well type is predicted, the normalized parameters of the five-point model are first loaded, and the corresponding input parameters and the volume prediction model are loaded. After the input parameters are processed by the volume prediction model, the predicted value is obtained. Then, the predicted value is collected and denormalized to obtain the true value, which is not specifically limited in the embodiment of the present application.

[0098] In another embodiment of the present application, for further definition and explanation, the steps further include:

[0099] Based on the volume prediction model, the well type, well spacing, permeability coefficient and pumping volume are predicted and processed to obtain fitting parameters of effective convection volume;

[0100] The fitting parameters are integrated with the fitting curve to obtain the predicted values ​​of effective convection volume at different time points.

[0101] In the embodiment of the present application, since the output parameters are the fitting parameters of the fitting curve during the model training, the fitting parameters include the optimal amplitude coefficient, slope factor, and offset obtained through training. Therefore, the fitting parameters are integrated with the fitting curve to obtain the effective convection volume prediction values ​​for different time points, such as Figure 2 shown.

[0102] In a specific implementation scenario, the process of training and predicting the effective convection volume prediction model includes:

[0103] 1. Initialize the parameter model. You can set 96 different well types, well spacing (the distance between injection wells and extraction wells), formation permeability, and single-hole extraction flow rate model parameters. Among them, the well types for in situ leaching uranium include five-point type and seven-point type, with 48 types of each.

[0104] 2. Call the open source program systems Modflow-2005 and Modpath7 to perform in situ leaching uranium flow field simulation and particle tracing simulation.

[0105] 3. Use the Log3P1 fitting tool in the existing software Origin and use the nonlinear least squares method to optimize the curve characteristic parameters a, b and c to maximize the regression coefficient of the model function and the simulated data and minimize the residual square sum. Finally, the curve characteristic parameter values ​​of the Log3P1 fitting results under different situations are obtained.

[0106] 4. Based on the a, b, and c parameters of the time-effective convection volume change curve under the known model parameter combination, determine the curve parameters of the neural network prediction of the unknown parameter combination. Among them, 38 categories of all cases of the five-point type (a total of 48 categories) and the seven-point type (a total of 48 categories) are selected as training sets for training the neural network model, and the remaining 10 categories are used as validation sets to verify the neural network model.

[0107] 5. According to step (3) and step (4), the five-point and seven-point neural network model training and prediction are performed respectively, and the predicted values ​​of the training set and the validation set are compared with the true values ​​respectively, and the prediction results of the parameters of the five-point model are obtained.

[0108] 6. Call the model corresponding to the five-point type or the model corresponding to the seven-point type to predict the a, b, and c parameters, and obtain the predicted value of the effective convection volume by reverse normalization.

[0109] The embodiment of the present application provides a method for predicting the effective convection volume of in-situ uranium leaching. Compared with the prior art, the embodiment of the present application obtains the well type, well spacing, permeability coefficient and pumping volume of in-situ uranium leaching; retrieves a volume prediction model corresponding to the well type, wherein the volume prediction model is obtained by training based on a constructed time volume training sample set, and the effective convection volume samples in the time volume training sample set are obtained by simulation based on injecting multiple tracking particles into a three-dimensional model; predicts and processes the well spacing, the permeability coefficient and the pumping volume based on the volume prediction model to obtain a predicted value of the effective convection volume, thereby reducing the prediction time of the effective convection volume as a leaching effect, achieving a more accurate prediction of the leaching effect, and meeting the high-precision requirements for in-situ uranium leaching under different mining environments, thereby greatly improving the output in actual generation.

[0110] Furthermore, as a response to the above Figure 1 The implementation of the method shown in the embodiment of the present application provides a prediction device for the effective convection volume of uranium in situ leaching, such as Figure 3 As shown, the device comprises:

[0111] An acquisition module 21 is used to obtain the well type, well spacing, permeability coefficient and pumping volume of in-situ uranium leaching;

[0112] A calling module 22 is used to call a volume prediction model corresponding to the well type, wherein the volume prediction model is obtained by training based on a constructed time volume training sample set, and the effective convection volume samples in the time volume training sample set are obtained by simulation based on injecting multiple tracking particles into a three-dimensional model;

[0113] The processing module 23 is used to perform prediction processing on the well spacing, the permeability coefficient and the pumping volume based on the volume prediction model to obtain a predicted value of the effective convection volume.

[0114] Furthermore, the device further comprises: a definition module, a simulation module,

[0115] The acquisition module is also used to acquire well spacing samples, permeability coefficient samples and pumping volume samples of different well types;

[0116] The definition module defines a three-dimensional model and injects multiple tracking particles into the three-dimensional model;

[0117] The simulation module is used to perform numerical simulation based on the multiple tracking particles and the well spacing samples, the permeability coefficient samples and the pumping volume samples to obtain an effective convection volume simulation value.

[0118] Furthermore, the device also includes:

[0119] A first construction module is used to construct a fitting curve based on the effective convection volume simulation value and the simulation time, wherein the fitting curve includes an amplitude coefficient, a tilt factor and an offset;

[0120] The second construction module is used to construct a time volume training sample set based on the optimally solved amplitude coefficient, slope factor, offset, and well spacing samples, permeability coefficient samples, and pumping volume samples of different well types.

[0121] Furthermore, the well type is a five-point type, and the device further comprises: a fourth construction module, a first training module,

[0122] The acquisition module is further used to acquire first fitting parameters of the first well spacing sample, the first permeability coefficient sample, the first pumping volume sample and the effective convection volume sample of the five-point type from the time volume training sample set;

[0123] The third building module is used to build a first multi-layer feedforward neural network;

[0124] The first training module is used to use the first well spacing sample, the first permeability coefficient sample, and the first pumping volume sample as first model input parameters, determine the first fitting parameter as the first model output parameter, and perform model training on the first multi-layer feedforward neural network to obtain the five-point first volume prediction model.

[0125] Furthermore, the well type is a seven-point type, and the device further comprises: a fifth construction module, a second training module,

[0126] The acquisition module is further used to acquire second fitting parameters of the second well spacing sample, the second permeability coefficient sample, the second pumping volume sample and the effective convection volume sample of the seven-point type from the time volume training sample set;

[0127] The fourth building block is used to build a second multi-layer feedforward neural network;

[0128] The second training module is used to use the second well spacing sample, the second permeability coefficient sample, and the second pumping volume sample as second model input parameters, determine the second fitting parameter as the second model output parameter, and perform model training on the second multi-layer feedforward neural network to obtain the five-point second volume prediction model.

[0129] Furthermore, the training module is also used to calculate the deviation during the training of the first multi-layer feedforward neural network or the second multi-layer feedforward neural network by using the mean square error as a loss function during the model training process, and complete the model training when the deviation is minimum; wherein the weight value of the first multi-layer feedforward neural network or the second multi-layer feedforward neural network is obtained by dynamically adjusting the sample loss value determined based on the callback function.

[0130] Further,

[0131] The processing module is also used to predict the well type, well spacing, permeability coefficient and pumping volume based on the volume prediction model to obtain fitting parameters of the effective convection volume; integrate the fitting parameters with the fitting curve to obtain the predicted values ​​of the effective convection volume at different time points.

[0132] The embodiment of the present application provides a prediction device for the effective convection volume of in-situ uranium leaching. Compared with the prior art, the embodiment of the present application obtains the well type, well spacing, permeability coefficient and pumping volume of in-situ uranium leaching; calls up a volume prediction model corresponding to the well type, wherein the volume prediction model is obtained by training based on a constructed time volume training sample set, and the effective convection volume samples in the time volume training sample set are obtained by simulation based on injecting multiple tracking particles into a three-dimensional model; based on the volume prediction model, the well spacing, the permeability coefficient and the pumping volume are predicted and processed to obtain a predicted value of the effective convection volume, thereby reducing the prediction time of the effective convection volume as the leaching effect, achieving a more accurate prediction of the leaching effect, and meeting the high-precision requirements for in-situ uranium leaching under different mining environments, thereby greatly improving the output in actual generation.

[0133] According to one embodiment of the present application, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the computer executable instruction can execute the method for predicting the effective convection volume of in-situ leaching uranium in any of the above method embodiments.

[0134] Figure 4 A schematic diagram of the structure of a terminal provided according to an embodiment of the present application is shown, and the specific embodiment of the present application does not limit the specific implementation of the terminal.

[0135] like Figure 4 As shown, the terminal may include: a processor (processor) 302 , a communication interface (Communications Interface) 304 , a memory (memory) 306 , and a communication bus 308 .

[0136] The processor 302 , the communication interface 304 , and the memory 306 communicate with each other via the communication bus 308 .

[0137] The communication interface 304 is used to communicate with other devices such as clients or other servers.

[0138] The processor 302 is used to execute the program 310, and specifically can execute the relevant steps in the above-mentioned embodiment of the method for predicting the effective convection volume of uranium in situ leaching.

[0139] Specifically, the program 310 may include program codes, which include computer operation instructions.

[0140] The processor 302 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in the terminal may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0141] The memory 306 is used to store the program 310. The memory 306 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0142] The program 310 may be specifically configured to enable the processor 302 to perform the following operations:

[0143] Obtain the well type, well spacing, permeability coefficient and pumping volume for in-situ leaching of uranium;

[0144] Retrieving a volume prediction model corresponding to the well type, wherein the volume prediction model is obtained by training based on a constructed time volume training sample set, and the effective convection volume samples in the time volume training sample set are obtained by simulation based on injecting multiple tracking particles into a three-dimensional model;

[0145] The well spacing, the permeability coefficient and the pumping volume are predicted based on the volume prediction model to obtain a predicted value of the effective convection volume.

[0146] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0147] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for predicting the effective convection volume of uranium in situ leaching, characterized in that: include: Obtain the well type, well spacing, permeability coefficient and pumping volume for in-situ leaching of uranium; Retrieving a volume prediction model corresponding to the well type, wherein the volume prediction model is obtained by training based on a constructed time volume training sample set, and the effective convection volume samples in the time volume training sample set are obtained by simulation based on injecting multiple tracking particles into a three-dimensional model; The well spacing, the permeability coefficient and the pumping volume are predicted based on the volume prediction model to obtain a predicted value of the effective convection volume.

2. The method according to claim 1, characterized in that Before obtaining the well type, well spacing, permeability coefficient and liquid extraction volume of in situ leaching uranium, the method further comprises: Obtain samples of well spacing, permeability coefficient, and pumping volume for different well types; defining a three-dimensional model, and injecting multiple tracking particles into the three-dimensional model; Numerical simulation is performed based on the multiple tracking particles and the well spacing samples, the permeability coefficient samples and the pumping volume samples to obtain an effective convection volume simulation value.

3. The method according to claim 2, characterized in that After performing numerical simulation based on the multiple tracking particles and the well spacing samples, the permeability coefficient samples, and the pumping volume samples to obtain the effective convection volume simulation value, the method further includes: Constructing a fitting curve based on the effective convection volume simulation value and the simulation time, wherein the fitting curve includes an amplitude coefficient, a tilt factor and an offset; A time-volume training sample set is constructed based on the optimally solved amplitude coefficient, slope factor, offset, and well spacing samples, permeability coefficient samples, and pumping volume samples of different well types.

4. The method according to claim 3, characterized in that The well type is a five-point type. Before calling the volume prediction model corresponding to the well type, the method further includes: Acquire first fitting parameters of the first well spacing sample, the first permeability coefficient sample, the first pumping volume sample and the effective convection volume sample of the five-point type from the time volume training sample set; Constructing the first multi-layer feedforward neural network; The first well spacing sample, the first permeability coefficient sample, and the first pumping volume sample are used as first model input parameters, the first fitting parameter is determined as a first model output parameter, and the first multi-layer feedforward neural network is trained to obtain the five-point first volume prediction model.

5. The method according to claim 3, characterized in that: The well type is a seven-point type. Before calling the volume prediction model corresponding to the well type, the method further includes: Acquire second fitting parameters of the second well spacing sample, the second permeability coefficient sample, the second pumping volume sample and the effective convection volume sample of the seven-point type from the time volume training sample set; Construct a second multi-layer feedforward neural network; The second well spacing sample, the second permeability coefficient sample, and the second pumping volume sample are used as second model input parameters, the second fitting parameter is determined as the second model output parameter, and the second multi-layer feedforward neural network is trained to obtain the seven-point second volume prediction model.

6. The method according to claim 4 or 5, characterized in that: The method further comprises: During the model training process, the deviation is calculated during the training process of the first multi-layer feedforward neural network or the second multi-layer feedforward neural network by using the mean square error as a loss function, and the model training is completed when the deviation is minimum; The weight value of the first multi-layer feedforward neural network or the second multi-layer feedforward neural network is obtained by dynamically adjusting the sample loss value determined based on the callback function.

7. The method according to any one of claims 1 to 6, characterized in that: The method further comprises: Based on the volume prediction model, the well type, well spacing, permeability coefficient and pumping volume are predicted and processed to obtain fitting parameters of effective convection volume; The fitting parameters are integrated with the fitting curve to obtain the predicted values ​​of effective convection volume at different time points.

8. A device for predicting the effective convection volume of uranium in situ leaching, characterized in that: include: An acquisition module is used to obtain the well type, well spacing, permeability coefficient and pumping volume of in-situ uranium leaching; A calling module, used for calling a volume prediction model corresponding to the well type, wherein the volume prediction model is obtained by training based on a constructed time volume training sample set, and the effective convection volume samples in the time volume training sample set are obtained by simulation based on injecting multiple tracking particles into a three-dimensional model; A processing module is used to predict the well spacing, the permeability coefficient and the pumping volume based on the volume prediction model to obtain a predicted value of the effective convection volume.

9. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to claim 1 are implemented.

10. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method of claim 1.

Images