Method and device for predicting porosity and permeability parameters of sandstone type uranium deposit

Abstract

The invention provides a method and a device for predicting porosity and permeability parameters of a sandstone type uranium mine layer, and relates to the field of in-situ leaching uranium mining, and the method comprises the steps: obtaining first logging data and first porosity and permeability parameters of a basic training area, and second logging data and second porosity and permeability parameters of a sandstone type uranium mine area; carrying out normalization processing on the first logging data and the first pore permeability parameter and then constructing a first data sample set; using the first data sample set to train the long and short term memory network model to obtain a basic prediction model; performing data enhancement processing on the second logging data and the second porosity and permeability parameters and then constructing a second data sample set; migrating different network layers of the basic prediction model to form a plurality of pre-selected models; and using the second data sample set to train the weights in the pre-selected models in sequence, and using the pre-selected model with the highest prediction precision as a final prediction model. The method can effectively solve the problems of porosity and permeability data sample scarcity and insufficient porosity and permeability parameter prediction precision of the sandstone type uranium mine area.

Description

Technical Field

[0001] The invention relates to the field of in-situ leaching of uranium, and in particular to a method and a device for predicting porosity and permeability parameters of a sandstone-type uranium ore layer. Background Art

[0002] The spatial distribution of porosity and permeability parameters of sandstone-type uranium deposits represents the migration and transformation of leachants and the leaching reaction process of uranium ore. At present, the geophysical parameters obtained through geophysical exploration (including natural potential, acoustic wave velocity, three-lateral resistivity, induced polarization, etc.) can be used to indicate the spatial distribution of porosity and permeability parameters. However, in addition to being affected by the porosity and permeability parameters of the ore layer, different types of geophysical parameters are also controlled by many other factors such as lithology, water content, and clay mineral content, resulting in a relatively complex relationship between geophysical data and porosity and permeability parameters, rather than a simple linear relationship. At present, there is no interpretation method that can achieve accurate conversion between geophysical data and porosity and permeability parameters. Summary of the invention

[0003] The purpose of the present invention is to provide a method and device for predicting porosity and permeability parameters of sandstone-type uranium ore layers to improve the above problems. In order to achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0004] In a first aspect, the present application provides a method for predicting porosity and permeability parameters of a sandstone-type uranium ore layer, comprising:

[0005] Acquire first well logging data and first porosity and permeability parameters of a basic training area and second well logging data and second porosity and permeability parameters of a sandstone-type uranium mining area, wherein the basic training area is an area with similar geological conditions to the sandstone-type uranium mining area;

[0006] After normalizing the first well logging data and the first porosity and permeability parameter, a first data sample set is constructed; a long short-term memory network model is constructed, and the long short-term memory network model is trained using the first data sample set to obtain a basic prediction model;

[0007] After performing data enhancement processing on the second well logging data and the second porosity and permeability parameter, a second data sample set is constructed according to a preset ratio;

[0008] Building an initial prediction model for porosity and permeability parameters of sandstone-type uranium mining areas, and migrating different network layers of the basic prediction model to the initial prediction model to obtain multiple pre-selected models;

[0009] The weights in each pre-selected model are trained in turn using the second data sample set, and the prediction accuracy of each pre-selected model is calculated, and the pre-selected model with the highest prediction accuracy is used as the final prediction model for the porosity and permeability parameters of the sandstone-type uranium mining area.

[0010] In a second aspect, the present application also provides a device for predicting porosity and permeability parameters of a sandstone-type uranium ore layer, comprising:

[0011] Acquisition module: acquiring first well logging data and first porosity and permeability parameters of a basic training area and second well logging data and second porosity and permeability parameters of a sandstone-type uranium mining area, wherein the basic training area is an area with similar geological conditions to the sandstone-type uranium mining area;

[0012] Model building module: constructing a first data sample set after normalizing the first well logging data and the first porosity and permeability parameter; building a long short-term memory network model, and training the long short-term memory network model using the first data sample set to obtain a basic prediction model;

[0013] Data enhancement module: after performing data enhancement processing on the second well logging data and the second porosity and permeability parameter, a second data sample set is constructed according to a preset ratio;

[0014] Migration module: building a porosity and permeability parameter prediction model for a target sandstone-type uranium mining area, and migrating different network layers of the basic prediction model to the porosity and permeability parameter prediction model for a sandstone-type uranium mining area to obtain multiple pre-selected models;

[0015] Training module: using the second data sample set to train each pre-selected model in turn, and calculating the prediction accuracy of each pre-selected model, and taking the pre-selected model with the highest prediction accuracy as the final prediction model of the porosity and permeability parameters.

[0016] In a third aspect, the present application also provides a device for predicting porosity and permeability parameters of a sandstone-type uranium ore layer, comprising:

[0017] Memory for storing computer programs;

[0018] A processor is used to implement the steps of the porosity and permeability parameter prediction method of the sandstone-type uranium ore layer when executing the computer program.

[0019] In a fourth aspect, the present application further provides a readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned method for predicting porosity and permeability parameters based on sandstone-type uranium ore layers are implemented.

[0020] The beneficial effects of the present invention are:

[0021] The present invention uses a large number of data samples under similar geological conditions to train a long short-term memory network model to generate a basic prediction model. At the same time, the present invention introduces a Gaussian noise algorithm and a k-means clustering algorithm to increase data samples for sandstone-type uranium mining areas to alleviate data imbalance. Then, the network layer of the basic prediction model is migrated to the target sandstone-type uranium mining area through transfer learning, and the model is locally trained and weighted using data samples from the target sandstone-type uranium mining area. Finally, a porosity and permeability parameter prediction model is obtained, which improves the accuracy and stability of the model.

[0022] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by implementing the embodiments of the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 It is a schematic flow chart of a method for predicting porosity and permeability parameters of a sandstone-type uranium ore layer described in an embodiment of the present invention;

[0025] Figure 2 is the RMSE during the LSTM model training process in the embodiment of the present invention;

[0026] Figure 3 is the MAE during the LSTM model training process in the embodiment of the present invention;

[0027] Figure 4 is the porosity relative error of the LSTM model in the embodiment of the present invention;

[0028] Figure 5 is the relative error of the permeability of the LSTM model in the embodiment of the present invention;

[0029] Figure 6 is the relative error of the porosity of each preselected model in the embodiment of the present invention;

[0030] Figure 7 is the relative error of the permeability of each preselected model in the embodiment of the present invention;

[0031] Figure 8 It is a schematic diagram of the structure of the porosity and permeability parameter prediction device of the sandstone type uranium ore layer described in the embodiment of the present invention;

[0032] Fig. 9 It is a schematic diagram of the structure of the porosity and permeability parameter prediction device for the sandstone-type uranium ore layer described in an embodiment of the present invention.

[0033] Markings in the figure:

[0034] 800. Porosity and permeability parameter prediction device for sandstone-type uranium ore layer; 801. Processor; 802. Memory; 803. Multimedia component; 804. I / O interface; 805. Communication component. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0036] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0037] Embodiment 1:

[0038] This embodiment provides a method for predicting porosity and permeability parameters of sandstone-type uranium deposits. Aiming at the problem that the amount of data obtained from core testing in sandstone-type uranium deposits that can directly calibrate the relationship between logging data and porosity and permeability parameters is too small, this method introduces a transfer learning algorithm. After training the long short-term memory network model using logging and porosity and permeability parameter data under other similar geological conditions such as oil fields, the model is fine-tuned locally using the data of the target mining area. This can effectively deal with the problem of too little data during model training and maintain the accuracy of predicting porosity and permeability parameters using logging data.

[0039] See also Figure 1 , the figure shows that the method includes:

[0040] S1. Acquire the first well logging data and the first porosity parameter of the basic training area and the second well logging data and the second porosity parameter of the sandstone-type uranium mining area, wherein the basic training area is an area with similar geological conditions to the sandstone-type uranium mining area;

[0041] Specifically, the step S1 includes:

[0042] S11. Collect the natural gamma (GR), well diameter (CAL), natural potential (SP), permeability and porosity corresponding to different depths in the basic training area. Some of the collected data are shown in Table 1:

[0043] Table 1

[0044] Depth(m) SP(mv) GR CAL Porosity Permeability 198 63 60 23 25 100 246 59 49 22 30 820 276 56 60 23 30 820 292 54 45 23 30 730 330 55 53 23 30 730 344 50 36 23 30 730 402 64 47 23 28 630 532 54 60 24 24 550 546 53 47 24 26 730

[0045] S12. The first logging data is composed of natural gamma, well diameter, and natural potential at each depth in the basic training area;

[0046] S13. The first porosity parameter is formed by the permeability and porosity corresponding to each depth in the basic training area;

[0047] S14. Collect the natural gamma, well diameter, natural potential, permeability and porosity corresponding to 9 different depths in the sandstone-type uranium deposit area. The 9 sets of data collected are shown in Table 2:

[0048] Table 2

[0049] Depth(m) SP GR CAL Porosity Permeability 37.7 -81.73 5.12 174.47 29.2 2282.8 38.4 -82.38 5.71 170.87 25.0 135.8 41.9 -89.06 5.53 155.88 7.2 10.1 47.1 -89.17 4.47 112.52 4.1 95.9 49.4 -86.89 4.72 121.37 11.5 5247.0 55.6 -92.59 6.58 120.01 26.1 2218.8 88.4 -70.75 4.31 147.26 28.7 910.4 93.5 -76.67 10.36 146.73 35.4 827.1 98.2 -80.6 10.12 139.81 22.2 218.7

[0050] S15. The second logging data is composed of natural gamma, well diameter and natural potential at each depth in the sandstone-type uranium deposit area;

[0051] S16. The permeability and porosity corresponding to each depth in the sandstone-type uranium mining area constitute a second porosity parameter, wherein the data volume of the first well logging data and the first porosity parameter is greater than the data volume of the second well logging data and the second porosity parameter.

[0052] Based on the above embodiments, the method further includes:

[0053] S2. constructing a first data sample set after normalizing the first logging data and the first porosity and permeability parameter; building a long short-term memory network model, and training the long short-term memory network model using the first data sample set to obtain a basic prediction model;

[0054] Specifically, due to the differences in dimensions and values ​​between different data, the influence of logging data with smaller values ​​on porosity parameters is easily ignored by the model, while the influence of logging data with higher values ​​on porosity parameters is easily overestimated, which will eventually affect the identification and prediction of porosity parameters. Therefore, it is necessary to standardize the logging data and scale the data proportionally so that indicators of different units or magnitudes can be compared and weighted. The standardization method used in this embodiment is the maximum and minimum normalization method, that is, deviation standardization, which maps the data to the range of [0,1]. The specific normalization formula is:

[0055]

[0056] Where: is the i-th data sample after normalization, x i is the ith data sample before normalization, i.e., natural gamma, wellbore, natural potential, permeability, or porosity, is the maximum value of the entire data set. is the minimum value of the entire data set, and n is the length of the data set. The logging data is processed by Min-Max standardization, and the values ​​are finally distributed in the interval (0,1).

[0057] Specifically, step S2 includes:

[0058] S21. Take the first porosity parameter of each depth as a set of output data In the formula, Indicates depth, represents the permeability, represents porosity, and i represents the i-th group of output data.

[0059] S22. The first well logging data of each depth and the first well logging data of the two adjacent heights are taken together as a set of input data. The set of input data can be uniformly written as:

[0060]

[0061] S23. A set of data samples is formed by input data and output data of the same depth to construct a first data sample set;

[0062] S24. Divide the first data sample set into a training set and a validation set in a ratio of 7:3.

[0063] Specifically, the step S2 further includes:

[0064] S25. A long short-term memory network model is constructed by an input layer, multiple long short-term memory (LSTM) network layers, a fully connected layer, and an output layer;

[0065] Specifically, the learning rate of the long short-term memory network model is set to 0.03, three LSTM layers are used, the number of hidden layer neural units is 8, the random dropout (Dropout) is 0.3, and the number of iterations is set to 600. Preferably, Dropout and early stopping mechanisms are introduced into the model to monitor the performance of the validation set and prevent overfitting of the model.

[0066] One of the long short-term memory (LSTM) network layers includes multiple LSTM memory cells. An LSTM cell consists of a forget gate, an input gate, an output gate, and a candidate memory cell. The forget gate is used to selectively forget the information in the memory cell state and retain useful information; the input gate controls the input of new information and selectively records it into the memory cell state; the output gate is used to ensure that the memory cell only outputs relevant information synchronized with the current time. The three gate structures ensure that information passes selectively by discarding useless information and adding useful information, so that the memory cell state is continuously updated, thereby achieving the purpose of controlling and protecting the memory cell state.

[0067] S26. Input the training set into the input layer, and organize each group of input data in the training set into a sequence;

[0068] S27. Input the multiple sequences into the long short-term memory network layer for feature extraction, and input the extracted features into the fully connected layer;

[0069] Specifically, the calculation process of extracting features of the long short-term memory network layer is:

[0070] Get the output h of the unit at time t-1 t-1 , according to h t-1 Calculate the result f of the forget gate t :

[0071] f t =σ(W f [h t-1 ,x t ]+b f ) (2)

[0072] In the formula, σ represents the sigmoid function, x t represents the input data at time t, b f Represents the bias term of the forget gate, W f Represents the weight matrix of the forget gate.

[0073] According to h t-1 Calculate the result of the input gate i t :

[0074] i t =σ(W i [h t-1 ,x t ]+b i ) (3)

[0075] Where W i represents the weight matrix of the input gate, b i Represents the bias term of the input gate.

[0076] Using h t-1 The result of getting the candidate value:

[0077]

[0078] In the formula, represents the candidate value, tanh represents the activation function, b c represents the bias term of the candidate memory unit, W c Represents the weight matrix of the candidate memory unit.

[0079] Using the cell state C at time t-1 t-1, update the cell states of the forget gate, input gate, and candidate values:

[0080]

[0081] In the formula, f t represents the output of the forget gate at time t, i t represents the input gate output at time t, C t Represents the cell state at time t.

[0082] The output gate result is calculated using the unit output at time t-1:

[0083] o t =σ(W o [h t-1 ,x t ]+b o ) (6)

[0084] In the formula, o t represents the output gate output at time t, W o represents the weight matrix of the output gate, b o Represents the bias term of the output gate.

[0085] Use the output gate and the cell state to calculate the output h of the cell at time t t :

[0086] h t =o t *tan h(C t ) (7)

[0087] In this embodiment, the forget gate will be the h output by the unit at the previous moment t-1 and x input at time t t At the same time, it is passed to the sigmoid function, which returns a number between 0 and 1. A number close to 0 means it should be discarded, and a number close to 1 means it should be retained. The input gate first converts the h output of the unit at the previous moment into t-1 The information and x input at time t t The sigmoid function and the tanh function are passed to the sigmoid function and the tanh function respectively. The sigmoid function has the same function as the forget gate, and the tanh function creates a new candidate state vector. The sigmoid function controls the part of the candidate state vector created by the tanh to be retained; the previous unit state c t-1 Multiply it with the output of the forget gate to decide whether the information should be discarded or retained, and then add the output of the input gate, that is, the product of the sigmoid and tanh calculation values, to get the updated unit state c t , the cell state is the core of the LSTM neural network; the h output of the cell at the previous moment t-1 and the x we ​​enter nowt Transmitted to the sigmoid function, and then the updated unit state c t Send it to the tanh function. Multiply the outputs of tanh and sigmoid to get the output h at the current moment t .

[0088] S28. The fully connected layer performs secondary extraction of input features for learning and training, and adjusts the parameters of the long short-term memory network model through the back propagation algorithm and the Adam optimizer;

[0089] In this embodiment, mean absolute error (MAE), root mean square error (RMSE) and relative error are used to evaluate the accuracy of the model.

[0090] During the training process, the model stopped training at 590 rounds, that is, when the performance on the validation set no longer improved, the training process was terminated in advance through the early stopping mechanism to avoid overfitting of the model on the training set and ensure that the model has good generalization ability on the validation set. After that, the RMSE of the training set and validation set gradually decreased and stabilized with the increase in the number of training rounds. Finally, the RMSE of the training set remained at around 0.1, and the RMSE of the validation set stabilized at around 0.12. Figure 2 shown.

[0091] During the training process, the MAE on the training set dropped rapidly from 0.5 to below 0.1, and the MAE on the validation set dropped from about 0.5 to about 0.15. As the number of training times increased, the MAE on the training set eventually stabilized at about 0.08, and the MAE on the validation set stabilized at about 0.09. Figure 3 shown.

[0092] In terms of porosity prediction, the relative error of porosity decreases with the iterations during the training process and finally stabilizes at 29%, reaching a prediction accuracy of 71%. Figure 4 shown.

[0093] In terms of permeability prediction, the relative error of permeability is stabilized at 25%, achieving a prediction accuracy of 75%, such as Figure 5 shown.

[0094] Based on the above embodiments, the method further includes:

[0095] S3. After performing data enhancement processing on the second well logging data and the second porosity and permeability parameters, a second data sample set is constructed according to a preset ratio;

[0096] Specifically, step S3 includes:

[0097] S31. Determine the noise standard deviation σ and mean μ, and generate a Gaussian noise array with the same data sample size as the second well logging data and the second porosity parameter according to the noise standard deviation σ, wherein the second well logging data may be natural gamma, well diameter and natural potential, and the second porosity parameter may be permeability or porosity;

[0098] Specifically, the Gaussian noise is a random process that follows a Gaussian (normal) distribution, and its probability density function takes a mean μ and a standard deviation σ as parameters to form a bell-shaped curve, and its probability density function is expressed as:

[0099]

[0100] Where p(x) represents the probability density of the random variable x at x;

[0101] A random number generator is used to generate random numbers, and the random numbers are transformed through normal distribution to make them obey standard normal distribution (ie, normal distribution with mean=μ and standard deviation=σ), and a Gaussian noise array is formed by the random numbers.

[0102] S32. Adding the Gaussian noise array to the second well logging data and the second porosity parameter to generate new second well logging data and the second porosity parameter;

[0103] This embodiment increases the diversity of data by adding Gaussian noise to the second logging data and the second porosity and permeability parameter, provides more data samples for the model, and helps to break the overfitting of the model to the training data, especially when the amount of data is limited.

[0104] S33. Determine the number of clusters required and randomly generate cluster centers for each cluster. In this embodiment, the objects in the data set are divided into K clusters, and the k-means clustering algorithm is used to divide the data samples so that the similarity of the members within the cluster is maximized and the similarity of the members between the clusters is minimized;

[0105] S34. Calculate the distance between each data sample and all cluster centers in the new second well logging data and the second porosity and permeability parameter in turn, and assign each data sample to the cluster to which the cluster center closest to it belongs;

[0106] S35. Calculate the average value of the data samples contained in each cluster, and use the average value to update the position of the cluster center;

[0107] S36. Re-dividing the data samples for each cluster using the updated cluster center until the cluster center of the cluster remains unchanged, indicating that the data sample division is completed;

[0108] S37. Add random noise to the cluster center to generate new data samples;

[0109] In this embodiment, the minority class samples are divided into multiple clusters through the k-means clustering algorithm, and then random noise is added to each cluster center to generate new minority class samples, thereby achieving data enhancement and balancing the category distribution in the data set, thereby generating a more representative and balanced data set, effectively improving the accuracy and robustness of the clustering process, significantly improving the quality of clustering and enhancing the adaptability of the model to new data.

[0110] Specifically, the step S3 further includes:

[0111] S31. The second porosity parameter of each depth is used as a set of output data In the formula, Indicates depth, represents the permeability, represents porosity, j represents the jth group of output data;

[0112] S32. Take the second well logging data of each depth and the second well logging data of two adjacent heights as a set of input data:

[0113]

[0114] S33. A set of data samples is composed of input data and output data of the same depth to construct a second data sample set;

[0115] Based on the above embodiments, the method further includes:

[0116] S4. Building an initial prediction model for porosity and permeability parameters of sandstone-type uranium mining areas, and migrating different network layers of the basic prediction model to the initial prediction model to obtain multiple pre-selected models;

[0117] In this embodiment, the first and third LSTM network layers (A), the first and second LSTM network layers (B), the first and second LSTM network layers (C), the third LSTM network layer (D), the second LSTM network layer (E), and the first LSTM network layer (F) are migrated respectively to obtain 6 pre-selected models, namely models A to F.

[0118] Based on the above embodiments, the method further includes:

[0119] S5. using the second data sample set to train the weights in each pre-selected model in turn, and calculating the prediction accuracy of each pre-selected model, and using the pre-selected model with the highest prediction accuracy as the final prediction model for porosity and permeability parameters of sandstone-type uranium deposits;

[0120] Specifically, step S5 includes:

[0121] S51. Divide the second data sample set into a first test group and a second test group;

[0122] S52. Inputting the first test group and the second test group into a preselected model respectively, and the preselected model outputs the corresponding first predicted porosity and permeability parameters and the second predicted porosity and permeability parameters;

[0123] S53. Calculate the first relative error between the first predicted porosity parameter and the second porosity parameter of the first test group, and the second relative error between the second predicted porosity parameter and the second porosity parameter of the second test group respectively;

[0124] S54. Calculate the mean error of the first relative error and the second relative error, and use the pre-selected model with the smallest mean error as the final prediction model, wherein the final prediction model is the model with the highest prediction accuracy;

[0125] See also Figure 6 The figure shows the porosity relative error comparison of the six pre-selected models. Model D has the best migration effect, with an error mean of 14% and a standard deviation of 22%.

[0126] See also Figure 7 ,The figure shows the relative error comparison of the permeability of the six pre-selected models. In the figure, model D has the best migration effect, with an error mean of 13% and a standard deviation of 20%;

[0127] From the above comparison data, it can be seen that the model prediction performance of the third LSTM network layer is the most stable, indicating the repeatability and high prediction accuracy of the model prediction results. Therefore, model D is used as the final prediction model for porosity parameters.

[0128] Embodiment 2:

[0129] like Figure 8 As shown, this embodiment provides a device for predicting porosity and permeability parameters of a sandstone-type uranium ore layer, the device comprising:

[0130] Acquisition module: acquiring first well logging data and first porosity and permeability parameters of a basic training area and second well logging data and second porosity and permeability parameters of a sandstone-type uranium mining area, wherein the basic training area is an area with similar geological conditions to the sandstone-type uranium mining area;

[0131] Model building module: constructing a first data sample set after normalizing the first well logging data and the first porosity and permeability parameter; building a long short-term memory network model, and training the long short-term memory network model using the first data sample set to obtain a basic prediction model;

[0132] Data enhancement module: after performing data enhancement processing on the second well logging data and the second porosity and permeability parameter, a second data sample set is constructed according to a preset ratio;

[0133] Migration module: building an initial prediction model for porosity and permeability parameters of sandstone-type uranium mining areas, and migrating different network layers of the basic prediction model to the initial prediction model to obtain multiple pre-selected models;

[0134] Training module: using the second data sample set to train the weights in each pre-selected model in turn, and calculating the prediction accuracy of each pre-selected model, and taking the pre-selected model with the highest prediction accuracy as the final prediction model for the porosity and permeability parameters of the sandstone-type uranium mining area.

[0135] Based on the above embodiment, the acquisition module includes:

[0136] The first acquisition unit: collects natural gamma, well diameter, natural potential, permeability and porosity corresponding to different depths in the basic training area;

[0137] The first component unit: the first well logging data is composed of natural gamma, well diameter and natural potential at each depth in the basic training area;

[0138] The second component unit: the first porosity parameter is composed of the permeability and porosity corresponding to each depth in the basic training area;

[0139] The second acquisition unit: collects the natural gamma, well diameter, natural potential, permeability and porosity corresponding to different depths in the sandstone-type uranium deposit area;

[0140] The third component unit: the second well logging data is composed of natural gamma, well diameter and natural potential at each depth in the sandstone-type uranium deposit area;

[0141] The fourth constituent unit: the second porosity parameter is constituted by the permeability and porosity corresponding to each depth of the sandstone-type uranium mining area, wherein the data volume of the first well logging data and the first porosity parameter is greater than the data volume of the second well logging data and the second porosity parameter.

[0142] Based on the above embodiments, the model building module includes:

[0143] The first assembly unit: taking the first porosity parameter of each depth as a set of output data;

[0144] The second forming unit: taking the first well logging data of each depth and the first well logging data of two adjacent heights as a group of input data;

[0145] The third building unit: forming a group of data samples from input data and output data of the same depth to construct a first data sample set;

[0146] Data division unit: divide the first data sample set into a training set and a validation set in a ratio of 7:3.

[0147] Based on the above embodiments, the model building module further includes:

[0148] Model building unit: The long short-term memory network model is built by the input layer, multiple long short-term memory network layers, the fully connected layer and the output layer;

[0149] Sequence organization unit: input the training set into the input layer and organize each set of input data in the training set into a sequence;

[0150] Feature extraction unit: input multiple sequences into the long short-term memory network layer for feature extraction, and input the extracted features into the fully connected layer;

[0151] Learning unit: The fully connected layer performs secondary extraction of input features for learning and training, and adjusts the parameters of the long short-term memory network model through the back-propagation algorithm and Adam optimizer.

[0152] Based on the above embodiment, the data enhancement module includes:

[0153] Noise generating unit: determining a noise standard deviation, and generating a Gaussian noise array having the same data sample size as the second well logging data and the second porosity parameter according to the noise standard deviation, wherein the second well logging data may be natural gamma, well diameter and natural potential, and the second porosity parameter may be permeability or porosity;

[0154] Noise adding unit: adding the Gaussian noise array to the second well logging data and the second porosity and permeability parameter to generate new second well logging data and second porosity and permeability parameter;

[0155] Random generation unit: determines the number of clusters required and randomly generates cluster centers for each cluster;

[0156] Sample allocation unit: sequentially calculates the distance between each data sample in the new second well logging data and the second porosity and permeability parameter and all cluster centers, and allocates each data sample to the cluster to which the cluster center closest to it belongs;

[0157] The first calculation unit calculates the average value of the data samples contained in each cluster, and updates the position of the cluster center using the average value;

[0158] The first division unit: re-dividing the data samples for each cluster using the updated cluster center until the cluster center of the cluster remains unchanged;

[0159] Sample generation unit: Add random noise to the cluster centers to generate new data samples.

[0160] Based on the above embodiment, the training module includes:

[0161] A second division unit: dividing the second data sample set after data enhancement into a first test group and a second test group;

[0162] Prediction unit: inputting the first test group and the second test group into the preselected model respectively, and the preselected model outputs the corresponding first predicted porosity and permeability parameters and the second predicted porosity and permeability parameters;

[0163] The second calculation unit is used to calculate the first relative error between the first predicted porosity parameter and the second porosity parameter of the first test group, and the second relative error between the second predicted porosity parameter and the second porosity parameter of the second test group;

[0164] The third calculation unit calculates the mean error of the first relative error and the second relative error, and takes the pre-selected model with the smallest mean error as the final prediction model, and the final prediction model is the model with the highest prediction accuracy.

[0165] It should be noted that, regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment of the method, and will not be elaborated here.

[0166] Embodiment 3:

[0167] Corresponding to the above method embodiment, this embodiment also provides a porosity and permeability parameter prediction device for a sandstone-type uranium ore layer. The porosity and permeability parameter prediction device for a sandstone-type uranium ore layer described below and the porosity and permeability parameter prediction method for a sandstone-type uranium ore layer described above can be referenced to each other.

[0168] Fig. 9 FIG. 8 is a block diagram of a porosity parameter prediction device 800 for a sandstone-type uranium ore layer according to an exemplary embodiment. Fig. 9 As shown, the porosity and permeability parameter prediction device 800 of the sandstone type uranium ore layer may include: a processor 801 and a memory 802. The porosity and permeability parameter prediction device 800 of the sandstone type uranium ore layer may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0169] The processor 801 is used to control the overall operation of the porosity parameter prediction device 800 of the sandstone-type uranium ore layer to complete all or part of the steps in the porosity parameter prediction method of the sandstone-type uranium ore layer. The memory 802 is used to store various types of data to support the operation of the porosity parameter prediction device 800 of the sandstone-type uranium ore layer, and these data may include, for example, instructions for any application or method operating on the porosity parameter prediction device 800 of the sandstone-type uranium ore layer, and application-related data, such as contact data, sent and received messages, pictures, audio, video, etc. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, disk or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 802 or sent via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, and the above-mentioned other interface modules can be keyboards, mice, buttons, etc. These buttons can be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the porosity parameter prediction device 800 of the sandstone-type uranium ore layer and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 805 may include: Wi-Fi module, Bluetooth module, NFC module.

[0170] In an exemplary embodiment, the porosity and permeability parameter prediction device 800 for a sandstone-type uranium ore layer can be implemented by one or more application-specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field programmable gate arrays (FPGA), controllers, microcontrollers, microprocessors or other electronic components to execute the above-mentioned porosity and permeability parameter prediction method for sandstone-type uranium ore layers.

[0171] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the steps of the porosity and permeability parameter prediction method of the sandstone-type uranium ore layer are implemented. For example, the computer-readable storage medium can be the memory 802 including the program instructions, and the program instructions can be executed by the processor 801 of the porosity and permeability parameter prediction device 800 of the sandstone-type uranium ore layer to complete the porosity and permeability parameter prediction method of the sandstone-type uranium ore layer.

[0172] Embodiment 4:

[0173] Corresponding to the above method embodiment, a readable storage medium is also provided in this embodiment. The readable storage medium described below and the porosity and permeability parameter prediction method of a sandstone-type uranium ore layer described above can be referred to each other.

[0174] A readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the porosity and permeability parameter prediction method of a sandstone-type uranium ore layer in the above method embodiment.

[0175] The readable storage medium may specifically be a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or other readable storage medium that can store program codes.

[0176] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0177] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for predicting porosity and permeability parameters of sandstone-type uranium deposits, characterized in that: include: Acquire first well logging data and first porosity and permeability parameters of a basic training area and second well logging data and second porosity and permeability parameters of a sandstone-type uranium mining area, wherein the basic training area is an area with similar geological conditions to the sandstone-type uranium mining area; Constructing a first data sample set after normalizing the first well logging data and the first porosity and permeability parameter; Building a long short-term memory network model, and using the first data sample set to train the long short-term memory network model to obtain a basic prediction model; After performing data enhancement processing on the second well logging data and the second porosity and permeability parameter, a second data sample set is constructed according to a preset ratio; Building an initial prediction model for porosity and permeability parameters of sandstone-type uranium mining areas, and migrating different network layers of the basic prediction model to the initial prediction model to obtain multiple pre-selected models; The weights in each pre-selected model are trained in turn using the second data sample set, and the prediction accuracy of each pre-selected model is calculated, and the pre-selected model with the highest prediction accuracy is used as the final prediction model for the porosity and permeability parameters of the sandstone-type uranium mining area.

2. The method for predicting porosity and permeability parameters of a sandstone-type uranium ore layer according to claim 1, characterized in that: Acquiring the first well logging data and the first porosity parameter of the basic training area and the second well logging data and the second porosity parameter of the sandstone type uranium deposit area, including: Collect natural gamma, well diameter, natural potential, permeability and porosity corresponding to different depths in the basic training area; The first logging data is composed of natural gamma, well diameter and natural potential at each depth in the basic training area; The first porosity parameter is formed by the permeability and porosity corresponding to each depth in the basic training area; Collect the natural gamma, well diameter, natural potential, permeability and porosity corresponding to different depths in sandstone-type uranium deposits; The second logging data is composed of natural gamma, well diameter and natural potential at each depth in the sandstone-type uranium deposit area; The permeability and porosity corresponding to each depth of the sandstone-type uranium mining area constitute the second porosity parameter, wherein the data volume of the first well logging data and the first porosity parameter is greater than the data volume of the second well logging data and the second porosity parameter.

3. The method for predicting porosity and permeability parameters of a sandstone-type uranium ore layer according to claim 1, characterized in that: After normalizing the first well logging data and the first porosity parameter, a first data sample set is constructed, including: The first porosity parameter of each depth is taken as a set of output data; The first well logging data of each depth and the first well logging data of two adjacent heights are taken together as a group of input data; A group of data samples is formed by input data and output data of the same depth to construct a first data sample set; The first data sample set is divided into a training set and a validation set in a ratio of 7:

3.

4. The method for predicting porosity and permeability parameters of a sandstone-type uranium ore layer according to claim 1, characterized in that: The second well logging data and the second porosity and permeability parameter are subjected to data enhancement processing, including: Determine a noise standard deviation, and generate a Gaussian noise array with the same data sample size as that of the second well logging data and the second porosity parameter according to the noise standard deviation, wherein the second well logging data may be natural gamma, well diameter and natural potential, and the second porosity parameter may be permeability or porosity; Adding the Gaussian noise array to the second well logging data and the second porosity and permeability parameters to generate new second well logging data and second porosity and permeability parameters; Determine the number of clusters required and randomly generate cluster centers for each cluster; The distance between each data sample and all cluster centers in the new second well logging data and the second porosity and permeability parameter is calculated in sequence, and each data sample is assigned to the cluster to which the cluster center closest to the data sample belongs; Calculate the average value of the data samples contained in each cluster, and use the average value to update the position of the cluster center; Use the updated cluster centers to re-divide the data samples for each cluster until the cluster centers remain unchanged; Add random noise to the cluster centers to generate new data samples.

5. The method for predicting porosity and permeability parameters of sandstone-type uranium ore layers according to claim 1, characterized in that: The weights in each pre-selected model are trained in turn by using the second data sample set, and the prediction accuracy of each pre-selected model is calculated, and the pre-selected model with the highest prediction accuracy is used as the final prediction model of the porosity and permeability parameters of the sandstone-type uranium mining area, including: Dividing the second data sample set equally into a first test group and a second test group; The first test group and the second test group are respectively input into the preselected model, and the preselected model outputs the corresponding first predicted porosity and permeability parameters and the second predicted porosity and permeability parameters; Calculate respectively a first relative error between the first predicted porosity parameter and the second porosity parameter of the first test group, and a second relative error between the second predicted porosity parameter and the second porosity parameter of the second test group; The mean error of the first relative error and the second relative error is calculated, and the pre-selected model with the smallest mean error is used as the final prediction model, and the final prediction model is the model with the highest prediction accuracy.

6. A device for predicting porosity and permeability parameters of sandstone-type uranium ore layers, characterized in that: include: Acquisition module: acquiring first well logging data and first porosity and permeability parameters of a basic training area and second well logging data and second porosity and permeability parameters of a sandstone-type uranium mining area, wherein the basic training area is an area with similar geological conditions to the sandstone-type uranium mining area; Model building module: constructing a first data sample set after normalizing the first well logging data and the first porosity and permeability parameter; Building a long short-term memory network model, and using the first data sample set to train the long short-term memory network model to obtain a basic prediction model; Data enhancement module: after performing data enhancement processing on the second well logging data and the second porosity and permeability parameter, a second data sample set is constructed according to a preset ratio; Migration module: building an initial prediction model for porosity and permeability parameters of sandstone-type uranium mining areas, and migrating different network layers of the basic prediction model to the initial prediction model to obtain multiple pre-selected models; Training module: using the second data sample set to train the weights in each pre-selected model in turn, and calculating the prediction accuracy of each pre-selected model, and taking the pre-selected model with the highest prediction accuracy as the final prediction model for the porosity and permeability parameters of the sandstone-type uranium mining area.

7. The porosity and permeability parameter prediction device for sandstone-type uranium ore layer according to claim 6, characterized in that: The acquisition module comprises: The first acquisition unit: collects natural gamma, well diameter, natural potential, permeability and porosity corresponding to different depths in the basic training area; The first component unit: the first well logging data is composed of natural gamma, well diameter and natural potential at each depth in the basic training area; The second component unit: the first porosity parameter is composed of the permeability and porosity corresponding to each depth in the basic training area; The second acquisition unit: collects the natural gamma, well diameter, natural potential, permeability and porosity corresponding to different depths in the sandstone-type uranium deposit area; The third component unit: the second well logging data is composed of natural gamma, well diameter and natural potential at each depth in the sandstone-type uranium deposit area; The fourth constituent unit: the permeability and porosity corresponding to each depth of the sandstone-type uranium mining area constitute the second porosity parameter, wherein the data volume of the first well logging data and the first porosity parameter is greater than the data volume of the second well logging data and the second porosity parameter.

8. The porosity and permeability parameter prediction device for sandstone-type uranium ore layer according to claim 6, characterized in that: The model building module includes: The first assembly unit: taking the first porosity parameter of each depth as a set of output data; The second forming unit: taking the first well logging data of each depth and the first well logging data of two adjacent heights as a group of input data; The third building unit: forming a group of data samples from input data and output data of the same depth to construct a first data sample set; Data division unit: divide the first data sample set into a training set and a validation set in a ratio of 7:

3.

9. The porosity and permeability parameter prediction device for sandstone-type uranium ore layer according to claim 6, characterized in that: The data enhancement module comprises: Noise generating unit: determining a noise standard deviation, and generating a Gaussian noise array having the same data sample size as the second well logging data and the second porosity parameter according to the noise standard deviation, wherein the second well logging data may be natural gamma, well diameter and natural potential, and the second porosity parameter may be permeability or porosity; Noise adding unit: adding the Gaussian noise array to the second well logging data and the second porosity and permeability parameter to generate new second well logging data and second porosity and permeability parameter; Random generation unit: determines the number of clusters required and randomly generates cluster centers for each cluster; Sample allocation unit: sequentially calculates the distance between each data sample in the new second well logging data and the second porosity and permeability parameter and all cluster centers, and allocates each data sample to the cluster to which the cluster center closest to it belongs; The first calculation unit calculates the average value of the data samples contained in each cluster, and updates the position of the cluster center using the average value; The first division unit: re-dividing the data samples for each cluster using the updated cluster center until the cluster center of the cluster remains unchanged; Sample generation unit: Add random noise to the cluster centers to generate new data samples.

10. The porosity and permeability parameter prediction device for sandstone-type uranium ore layers according to claim 6, characterized in that: The training module includes: A second division unit: dividing the second data sample set after data enhancement into a first test group and a second test group; Prediction unit: inputting the first test group and the second test group into the preselected model respectively, and the preselected model outputs the corresponding first predicted porosity and permeability parameters and the second predicted porosity and permeability parameters; The second calculation unit is used to calculate the first relative error between the first predicted porosity parameter and the second porosity parameter of the first test group, and the second relative error between the second predicted porosity parameter and the second porosity parameter of the second test group; The third calculation unit calculates the mean error of the first relative error and the second relative error, and takes the pre-selected model with the smallest mean error as the final prediction model, and the final prediction model is the model with the highest prediction accuracy.

Images