Mineral corrosion parameter prediction model construction method, prediction method, device and computer equipment

Abstract

The invention provides a mineral corrosion parameter prediction model construction method, a prediction method, a prediction device and computer equipment, and belongs to the field. According to the method, the mineral dissolution parameter prediction initial model is obtained according to the total dimension training of the mineral dissolution action sample and the mineral dissolution condition parameters, the mineral dissolution parameter prediction initial model is trained by utilizing the mineral dissolution parameter action sample, and the mineral dissolution parameter prediction model is obtained, so that the problem that the reservoir stratum has a long-term mineral dissolution time of millions of years is solved. The problems that due to many influence factors, the calculation amount of mineral corrosion parameter simulation prediction is large, uncertainty is high, errors are large, the reservoir evaluation precision is low, and oil-gas exploration is limited are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of reservoir parameter prediction, and in particular to a method for constructing a mineral dissolution parameter prediction model, a prediction method, a device and a computer device. Background Art

[0002] Mineral dissolution mostly occurs in geological environments within several thousand meters underground. The mechanism and quality evaluation of reservoir mineral dissolution are the key technical bottlenecks faced by today's oil and gas exploration and development, and the restoration of the evolution process of reservoir mineral dissolution is the core scientific problem that needs to be solved urgently. With the development of mineral dissolution research, the numerical simulation technology that has emerged in recent years can restore the evolution process of mineral dissolution, and to a certain extent, can realize the quantitative evaluation of mineral dissolution parameters such as mineral content, ion concentration, reservoir porosity, and permeability in mineral dissolution. Take the Mahu Sag as an example. The Fengcheng Formation in the Mahu Sag is rich in alkaline minerals, among which boraxite appears densely in certain layers. The organic-rich mudstone of the Fengcheng Formation is the main source rock, and the organic acid produced is very easy to react with the boraxite layer, thereby affecting the development of the reservoir. However, because the reservoir mineral dissolution time is as long as one million years, there are as many as hundreds of influencing factors, which results in large amount of calculation, strong uncertainty, and strong human operation in the simulation of mineral dissolution parameters, resulting in low reservoir evaluation accuracy, which limits oil and gas exploration. Summary of the invention

[0003] In view of the problems of the prior art, the present invention aims to provide a method for constructing a mineral dissolution parameter prediction model, a prediction method, an apparatus and a computer device to solve the problems of high uncertainty, large human factors and low accuracy in mineral dissolution simulation in the prior art.

[0004] In a first aspect, the present invention provides a method for constructing a mineral dissolution parameter prediction model, comprising the following steps:

[0005] Acquire a plurality of mineral dissolution parameter action samples, wherein the plurality of mineral dissolution parameter action samples all include mineral dissolution condition parameters and actual mineral dissolution parameters obtained according to the evolution of the mineral dissolution condition parameters;

[0006] Constructing an initial model for predicting mineral dissolution parameters according to the mineral dissolution sample and the total dimension of the mineral dissolution condition parameters;

[0007] The mineral dissolution parameter prediction initial model is trained using the mineral dissolution parameter action sample until the error between the mineral dissolution parameter prediction value obtained by the mineral dissolution parameter prediction initial model and the actual mineral dissolution parameter is within a preset error range or the mineral dissolution parameter prediction value reaches a preset accuracy rate, thereby obtaining a mineral dissolution parameter prediction model.

[0008] Preferably, before using the mineral dissolution parameter action sample to train the initial model for predicting the mineral dissolution parameters, it also includes: performing feature selection on the mineral dissolution condition parameters, and removing the mineral dissolution condition parameters whose influence coefficient on the actual mineral dissolution parameters is less than a preset value.

[0009] Preferably, according to the mineral dissolution sample and the total dimension of the mineral dissolution condition parameters, an initial model for predicting mineral dissolution parameters is constructed, which specifically includes the following steps:

[0010] When the total dimension of the mineral dissolution condition parameters is less than a preset dimension threshold, a machine learning model is constructed based on the mineral dissolution sample, and the machine learning model is used as an initial model for predicting mineral dissolution parameters; the algorithm adopted by the machine learning model includes at least one of a decision tree algorithm, a naive Bayes algorithm, a support vector machine algorithm, a random forest algorithm, an artificial neural network algorithm, a Boosting and Bagging algorithm, an association rule algorithm, and an expectation-maximization algorithm.

[0011] When the total dimension of the mineral dissolution condition parameters is greater than or equal to a preset dimension threshold, a deep learning network model is constructed according to the mineral dissolution sample, and the deep learning network model is used as an initial model for predicting mineral dissolution parameters.

[0012] Preferably, after obtaining a plurality of mineral dissolution parameter action samples and before using the mineral dissolution parameter action samples to train the mineral dissolution parameter prediction initial model, the method further includes:

[0013] The mineral dissolution samples are divided into training sets and test sets according to a preset ratio by using a random sampling method or a stratified sampling method;

[0014] The following formula is used to normalize the mineral dissolution condition parameters in the training set:

[0015]

[0016] Among them, x i is the mineral dissolution condition parameter of the i-th dimension in the training set, x i ′ is x i After normalization, the value of i ranges from 1 to n, where n is the dimension of the mineral dissolution condition parameter; μ i is the mean value of the mineral dissolution condition parameter in the i-th dimension, δ i is the standard deviation of the mineral dissolution condition parameters in the i-th dimension.

[0017] Preferably, the dissolution condition parameters include a mineral dissolution prediction period, and the dissolution condition parameters also include one or a combination of several of ion concentration, mineral content, temperature and pressure conditions, acidity and alkalinity, and porosity.

[0018] Preferably, the actual dissolution parameters include at least one or a combination of ion concentration, mineral content, temperature and pressure conditions, acidity and alkalinity, and porosity after the evolution time reaches the predicted period of mineral dissolution.

[0019] Preferably, the dissolution condition parameter values ​​in the mineral dissolution second parameter set in the mineral dissolution condition parameters are measured values ​​obtained at a certain observation time, or are measured values ​​obtained at multiple observation times.

[0020] In a second aspect, the present invention also provides a method for predicting mineral dissolution parameters, comprising the following steps:

[0021] Collect mineral dissolution condition parameters;

[0022] The collected mineral dissolution condition parameters are input into the mineral dissolution parameter prediction model obtained by the above-mentioned mineral dissolution parameter prediction model construction method to obtain the predicted mineral dissolution parameters.

[0023] In a third aspect, the present invention further provides a device for constructing a mineral dissolution parameter prediction model, comprising:

[0024] An acquisition module, used for acquiring a plurality of mineral dissolution parameter action samples, wherein the plurality of mineral dissolution parameter action samples all include mineral dissolution condition parameters and actual mineral dissolution parameters obtained according to the evolution of the mineral dissolution condition parameters;

[0025] A construction module, for constructing an initial model for predicting mineral dissolution parameters according to the mineral dissolution sample and the total dimension of the mineral dissolution condition parameters;

[0026] A training module is used to train the initial model for predicting mineral dissolution parameters using the mineral dissolution parameter action sample, until the error between the predicted value of the mineral dissolution parameter obtained by the initial model for predicting mineral dissolution parameters and the actual mineral dissolution parameter is within a preset error range or the predicted value of the mineral dissolution parameter reaches a preset accuracy, thereby obtaining a mineral dissolution parameter prediction model.

[0027] In a fourth aspect, the present invention further provides a mineral dissolution parameter prediction device, comprising:

[0028] Collection module, used to collect mineral dissolution condition parameters;

[0029] The prediction module is used to input the mineral dissolution condition parameters into the mineral dissolution parameter prediction model obtained by the above-mentioned mineral dissolution parameter prediction model construction device to obtain the mineral dissolution parameters predicted according to the mineral dissolution condition parameters.

[0030] In a fifth aspect, the present invention further provides a computer device, comprising a processor, which can execute the above-mentioned method for constructing a mineral dissolution parameter prediction model when running.

[0031] Compared with the prior art, the present invention has at least the following beneficial effects:

[0032] The present invention obtains an initial model for predicting mineral dissolution parameters based on the total dimension training of mineral dissolution samples and mineral dissolution condition parameters, and uses mineral dissolution parameter action samples to train the initial model for predicting mineral dissolution parameters to obtain a mineral dissolution parameter prediction model, which solves the problem that the simulation prediction of mineral dissolution parameters has a large amount of calculation, strong uncertainty and large error due to the fact that the reservoir mineral dissolution time is as long as one million years and there are many influencing factors, which in turn leads to low reservoir evaluation accuracy and limits oil and gas exploration. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A flow chart showing a method for constructing a mineral dissolution parameter prediction model provided by an embodiment of the present invention;

[0034] Figure 2 A schematic diagram showing a comparison between a predicted value of a mineral dissolution parameter and an actual mineral dissolution parameter according to an embodiment of the present invention is shown;

[0035] Figure 3 A schematic diagram showing the structure of a deep learning network model constructed in one embodiment of the present invention is shown;

[0036] Figure 4 A schematic diagram of the structure of a deep learning network model after being processed by a temporary retreat method in one embodiment of the present invention is shown;

[0037] Figure 5 A flow chart of a method for predicting mineral dissolution parameters provided by an embodiment of the present invention is shown;

[0038] Figure 6 A schematic diagram of the structure of a device for constructing a mineral dissolution parameter prediction model provided by an embodiment of the present invention is shown;

[0039] Figure 7 A schematic diagram showing the structure of a mineral dissolution parameter prediction device provided by an embodiment of the present invention is shown;

[0040] Figure 8 A schematic structural diagram of a computer device provided by an embodiment of the present invention is shown.

[0041] In the figure:

[0042] 61. Get module;

[0043] 62. Building blocks;

[0044] 63. Training module;

[0045] 71. Collection module;

[0046] 72. Prediction module;

[0047] 802. Computer equipment;

[0048] 804, processor;

[0049] 806. Memory;

[0050] 808, driving mechanism;

[0051] 810, input / output module;

[0052] 812. Input device;

[0053] 814. Output device;

[0054] 816. Presentation equipment;

[0055] 818. Graphical user interface;

[0056] 820, network interface;

[0057] 822, communication link;

[0058] 824. Communication bus. DETAILED DESCRIPTION

[0059] The specific implementation modes of the present invention are described in detail below in conjunction with the accompanying drawings.

[0060] Mineral dissolution refers to the process of converting loose sediments into sedimentary rocks under certain pressure and temperature. The dissolution of reservoir minerals takes millions of years and is affected by hundreds of factors. The existing numerical simulation technology of mineral dissolution has the problems of large amount of calculation, strong uncertainty and large error.

[0061] In order to solve the above problems, the embodiments of the present invention provide a method for constructing a mineral dissolution parameter prediction model, a prediction method, an apparatus and a computer device, which can be used to predict mineral dissolution parameters in mineral dissolution.

[0062] Figure 1It is a flow chart of a method for constructing a mineral dissolution parameter prediction model provided by an embodiment of the present invention. The method operation steps described in the embodiment or flow chart are provided here, but more or fewer operation steps may be included based on conventional or non-creative labor. The order of steps listed in the embodiment is only one way of executing the steps among many orders, and does not represent the only order of execution. When the system or device product is executed in practice, it can be executed in the order of the method shown in the embodiment or the accompanying drawings or in parallel. Specifically, Figure 1 As shown in FIG. 1 , the method for constructing a mineral dissolution parameter prediction model includes:

[0063] S110: Acquire a plurality of mineral dissolution parameter action samples, wherein the plurality of mineral dissolution parameter action samples all include mineral dissolution condition parameters and actual mineral dissolution parameters obtained according to the evolution of the mineral dissolution condition parameters.

[0064] The mineral dissolution condition parameters include the mineral dissolution prediction period, and the mineral dissolution condition parameters also include at least one or a combination of ion concentration, mineral content, temperature and pressure conditions, acidity and alkalinity, and porosity. The mineral dissolution condition parameters may also include other parameters not included, for example, they may also include geological condition parameters of the reservoir sample, such as longitude and latitude, burial depth, etc. The actual mineral dissolution parameters include at least one or a combination of ion concentration, mineral content, temperature and pressure conditions, acidity and alkalinity, and porosity after the evolution time reaches the mineral dissolution prediction period. Of course, the actual mineral dissolution parameters may also include other parameters not included.

[0065] For example, the mineral dissolution condition parameters may be the mineral content (e.g., the ratio of calcite and dolomite contents) and the mineral dissolution prediction period (e.g., three years), and the actual mineral dissolution parameters may be the mineral content measured in the reservoir after three years (i.e., the ratio of calcite and dolomite contents after three years).

[0066] For another example, the mineral dissolution condition parameters may also be mineral content (for example, the ratio of calcite and dolomite), ion concentration, temperature and pressure conditions, and mineral dissolution prediction period (for example, three years), while the actual mineral dissolution parameter may be the porosity of the reservoir three years later (or some other parameters that are difficult to collect or difficult to collect in actual geology).

[0067] That is, the actual mineral dissolution parameters can be the same type of parameters as the mineral dissolution condition parameters, or they can be different types of parameters. Therefore, different mineral dissolution parameter prediction models can be trained according to different mineral dissolution samples and used to predict mineral dissolution parameters in different application scenarios, with a wide range of applications.

[0068] It should be noted that for multiple mineral dissolution samples used in the same mineral dissolution parameter prediction scenario, the mineral dissolution prediction period in their mineral dissolution condition parameters should be the same. That is, if in multiple mineral dissolution samples, the mineral dissolution condition parameters are the mineral content and the three-year mineral dissolution prediction period, the actual mineral dissolution parameter used for prediction is the mineral content of the reservoir three years later; and if the mineral dissolution condition parameters are the mineral content and the one-year mineral dissolution prediction period, the actual mineral dissolution parameter used for prediction is the mineral content of the reservoir one year later, due to the different mineral dissolution prediction periods, they correspond to two specific mineral dissolution parameter prediction scenarios.

[0069] In the above two examples, the parameters of the mineral dissolution condition except the mineral dissolution prediction period are the measured values ​​obtained at a certain observation time (ie, before the three-year mineral dissolution prediction period of the reservoir).

[0070] However, if the parameters of the mineral dissolution condition except the mineral dissolution prediction period are all measured values ​​obtained at a certain observation time, it cannot reflect the complex process of actual reservoir evolution and is inconsistent with the actual mineral dissolution process, making the simulation of mineral dissolution too linear.

[0071] Therefore, in some preferred embodiments, the ion concentration, mineral content, temperature and pressure conditions, acidity and alkalinity, and porosity included in the mineral dissolution condition parameters can also be measured values ​​obtained at multiple observation times. Exemplarily, for a mineral dissolution sample of a certain reservoir, the ion concentration (as well as mineral content, temperature and pressure conditions, burial depth, etc.) of the reservoir can be measured once at different observation times within a three-year mineral dissolution prediction cycle. It should be noted that the time interval between two adjacent observation times may be equal or unequal; when the time interval between any two observation times is equal, it is periodic sampling, for example, within a three-year mineral dissolution prediction cycle, the ion concentration of the reservoir sample is obtained every 100 days. These discrete parameters are then obtained as inputs to the initial model for predicting mineral dissolution parameters. In addition to the mineral dissolution prediction cycle, the mineral dissolution condition parameters of other dimensions can also have multiple sub-dimensions (the measurement value obtained at one observation time corresponds to one sub-dimension).

[0072] S120: Constructing an initial model for predicting mineral dissolution parameters according to the mineral dissolution sample and the total dimension of the mineral dissolution condition parameters.

[0073] In one embodiment of the present invention, the total dimension of the mineral dissolution condition parameter is related to the number of dimensions of the mineral dissolution condition parameter (denoted as n), and the number of sub-dimensions of each mineral dissolution condition parameter other than the mineral dissolution prediction period. Exemplarily, when the mineral dissolution condition parameter is the mineral dissolution prediction period, mineral content and ion concentration, and the mineral content and ion concentration are both measured values ​​obtained at a certain observation time, its dimension is 3, the sub-dimension of each dimension of the mineral dissolution condition parameter except the mineral dissolution prediction period is 1, and the total dimension is 3. When the mineral dissolution condition parameter includes the mineral dissolution prediction period, mineral content and ion concentration, and the mineral content and ion concentration are both measured values ​​obtained at multiple (for example, m) observation times, the dimension is 3, the sub-dimensions of the mineral content and the ion concentration are both m, and the total dimension is 2m+1. When the mineral dissolution condition parameters include mineral dissolution prediction period, mineral content and ion concentration, where the mineral content is the measurement value obtained at a certain observation time, and the ion concentration is the measurement value obtained at multiple (for example, m) observation times, then the dimension is 3, the sub-dimension of the mineral content is 1, the sub-dimension of the ion concentration is m, and the total dimension is m+2.

[0074] It can be understood that when the ion concentration, mineral content, temperature and pressure conditions, acidity and alkalinity, and porosity in the mineral dissolution condition parameters are the total dimensions of the measurement values ​​obtained at multiple observation times, compared with the measurement values ​​of the ion concentration, mineral content, temperature and pressure conditions, acidity and alkalinity, porosity, etc. in the mineral dissolution condition parameters obtained at a certain observation time, the increase in the total number of dimensions is beneficial to improving the richness of mineral dissolution samples, thereby helping to improve the accuracy of the trained mineral dissolution parameter prediction model and improve the accuracy of the mineral dissolution parameters predicted subsequently using the model.

[0075] S130: Using the mineral dissolution parameter action sample to train the initial model for predicting mineral dissolution parameters, until the error between the predicted value of the mineral dissolution parameters obtained by the initial model for predicting mineral dissolution parameters and the actual mineral dissolution parameters is within a preset error range or the predicted value of the mineral dissolution parameters reaches a preset accuracy, thereby obtaining a mineral dissolution parameter prediction model.

[0076] Among them, the preset accuracy can be set to 90%, 95%, etc., and the accuracy of the predicted value of the mineral dissolution parameter can be calculated during the test set test.

[0077] like Figure 2 FIG. 1 is a schematic diagram showing a comparison between the predicted values ​​of mineral dissolution parameters and the actual mineral dissolution parameters according to an embodiment of the present invention. Specifically, the ordinate in the figure is porosity, and the abscissa is the sample number, that is, there are 200 mineral dissolution samples, each of which includes mineral dissolution condition parameters and actual mineral dissolution parameters, so there are 200 actual mineral dissolution parameters (i.e., corresponding to Figure 2 The upper part of the true value is the actual mineral dissolution parameter corresponding to each sample number).

[0078] The mineral dissolution condition parameters of each mineral dissolution sample are input into the mineral dissolution parameter prediction initial model, and the mineral dissolution parameter prediction initial model will output the corresponding mineral dissolution parameter prediction value, that is, there are a total of 200 mineral dissolution parameter prediction values ​​(that is, corresponding to Figure 2 In the lower part, predictvalue is the predicted value of mineral dissolution parameters corresponding to each sample number).

[0079] When the error between the predicted value of the mineral dissolution parameter and the actual mineral dissolution parameter is less than the preset error range or the predicted value of the mineral dissolution parameter reaches the preset accuracy, the training of the initial model for predicting the mineral dissolution parameter is completed, and the model obtained after the training is the mineral dissolution parameter prediction module. After calculating the error between 200 sets of actual mineral dissolution parameters and the corresponding predicted values ​​of mineral dissolution parameters, it can be known whether the initial model for predicting the mineral dissolution parameters is successfully trained.

[0080] An embodiment of the present invention provides a method for constructing a mineral dissolution parameter prediction model. An initial model for predicting mineral dissolution parameters is obtained by training based on mineral dissolution samples and the total dimension of mineral dissolution condition parameters. A mineral dissolution parameter prediction model is obtained by training based on existing mineral dissolution samples. The model is applied to various prediction scenarios, such as predicting current data of one or more parameters based on historical data of the one or more parameters, and predicting other parameters that are difficult to obtain in actual geology based on historical data of one or more parameters. This method solves the problem that the simulation prediction of mineral dissolution parameters has a large amount of calculation, strong uncertainty, and large errors due to the fact that the reservoir mineral dissolution time is as long as millions of years and there are many influencing factors, which in turn leads to low reservoir evaluation accuracy and limits oil and gas exploration.

[0081] Since there may be a large number of mineral dissolution condition parameters in the mineral dissolution sample, some of which may be irrelevant to the actual mineral dissolution parameters, in order to avoid the curse of dimensionality and reduce the difficulty of subsequent training of the initial model for mineral dissolution prediction to obtain the mineral dissolution parameter prediction model, in one embodiment of the present invention, in step S130, before using the mineral dissolution parameter action sample to train the initial model for mineral dissolution parameter prediction, it also includes: performing feature selection on the mineral dissolution condition parameters, and removing the mineral dissolution condition parameters whose influence coefficient on the actual mineral dissolution parameters is less than a preset value.

[0082] Specifically, the characteristic selection of mineral dissolution condition parameters can be achieved through filtering selection, wrapping selection, embedded selection and other methods:

[0083] A typical example of filter selection is the Relief method. The Relief algorithm is a feature weighting algorithm, which is implemented as follows:

[0084] A mineral dissolution parameter R is randomly selected from the mineral dissolution sample, and then the nearest neighbor parameter H is found from the parameters of the same type as parameter R (called the guessed neighbor, NearHit); and the nearest neighbor parameter M is found from the parameters of a different type from parameter R (called the guessed wrong neighbor, NearMiss).

[0085] Then update the initial weight of each parameter according to the following rules: if the distance between parameter R and parameter H in a certain feature is smaller than the distance between parameter R and parameter M in a certain feature, it means that the certain feature is useful for distinguishing the nearest neighbors of the same and different classes, and the weight of the feature is increased; conversely, if the distance between parameter R and parameter H in a certain feature is larger than the distance between parameter R and parameter M, it means that the feature has a negative effect on distinguishing the nearest neighbors of the same and different classes, and the weight of the feature is reduced.

[0086] Repeat the above process several times, and finally get the average weight of each feature. The larger the weight of a feature, the stronger its classification ability; conversely, the weaker its classification ability.

[0087] The features with weights lower than a preset value are removed to obtain the mineral dissolution condition parameters after feature selection.

[0088] The filtering selection method is to first use the feature selection process to "filter" the initial parameter features in the mineral dissolution sample, and then use the filtered parameters to train the subsequent model.

[0089] Wrapper selection does not consider the differences of subsequent models. Wrapper feature selection directly uses the performance of the model to be used in the end as the evaluation criterion for the feature subset. In other words, the purpose of wrapper feature selection is to select a tailor-made feature subset that is most conducive to the performance of a given model. Typical methods of wrapper selection include the LVW method (LasVegasWrapper method).

[0090] Embedded feature selection integrates the feature selection process with the model training process. The two are optimized in the same optimization process, that is, feature selection is automatically performed during the model training process. For example, L1 regularization, decision tree learning, etc.

[0091] Through filtering selection, wrapping selection, embedded selection and other methods, irrelevant, redundant parameters that have no ability to differently characterize the actual mineral dissolution parameters can be eliminated, and optimized mineral dissolution condition parameters can be obtained and used for model training, avoiding the dimensionality curse and reducing the difficulty of training the initial model for mineral dissolution prediction.

[0092] Furthermore, in one embodiment of the present invention, step S120: constructing an initial model for predicting mineral dissolution parameters according to the mineral dissolution sample and the total dimension of the mineral dissolution condition parameters may further include:

[0093] When the total dimension of the mineral dissolution condition parameters is less than a preset dimension threshold, a machine learning model is constructed according to the mineral dissolution sample, and the machine learning model is used as an initial model for predicting mineral dissolution parameters;

[0094] When the total dimension of the mineral dissolution condition parameters is greater than or equal to a preset dimension threshold, a deep learning network model is constructed according to the mineral dissolution sample, and the deep learning network model is used as an initial model for predicting mineral dissolution parameters.

[0095] The machine learning model is suitable for prediction when the total number of parameter dimensions is small, while the deep learning network model is suitable for prediction when the total number of parameter dimensions is large. Therefore, in one embodiment of the present invention, the type of model to be constructed can be selected according to the number of mineral dissolution condition parameter dimensions and the number of mineral dissolution condition parameter sub-dimensions in actual application, and finally a suitable model is selected to improve the mineral dissolution prediction effect while reducing the cost of model construction and model training.

[0096] Furthermore, in one embodiment of the present invention, a machine learning model can be constructed based on a mineral dissolution sample using a BP neural network algorithm optimized by FA, ​​and the machine learning model is used as an initial model for predicting mineral dissolution parameters, and the predicted values ​​of mineral dissolution parameters obtained by the machine learning model are converged with the loss function constructed by the actual mineral dissolution parameters, so as to train and obtain the mineral dissolution parameter prediction model. Of course, in addition to the BP neural network algorithm optimized by FA, ​​machine learning models can also be constructed using methods such as GBDT and XGBoost.

[0097] BP neural network is a multi-layer feedforward network trained by the error back propagation algorithm. It uses the gradient descent and mean square error method to modify the network connection weights. It is widely used and is one of the mature neural network algorithms. The three-layer BP neural network can meet the accuracy requirements of the design and achieve the approximation of any nonlinear function. Let the output of the hidden layer be Y j ; The threshold of the hidden layer is θ j ; The threshold of the output layer is θ k; The transfer function of the hidden layer is f1; The transfer function of the output layer is f2; The weight from the input layer to the hidden layer is w ij ; The weight from the hidden layer to the output layer is W jk . Input a learning sample (x k , b k ), where k∈{1, 2…N}, N is the number of samples, and the output Y of the jth neuron in the hidden layer is obtained j , network output Y k , T k For the expected output

[0098]

[0099] According to the principle of dynamic reverse error propagation algorithm, the steepest descent method is used to adjust the connection weights so that the error function E reaches the minimum value at the fastest speed. The basic formula for weight correction is:

[0100]

[0101] Where η is the learning rate and w is the weight. Network training is a process of continuous readjustment between weights and thresholds, with the goal of reducing network errors to a preset minimum value or stopping at a preset training step. Finally, the test sample is input into the trained network to obtain the result.

[0102] The use of FA to optimize BP neural network mainly utilizes FA's ability of fast iteration and global optimization to optimize the initial weights and thresholds of BP network, improve network learning performance, and avoid falling into local optimal solutions. The process of FA optimization BP neural network (FABP) algorithm is as follows Figure 1 Shown

[0103] ① It is known that the number of neurons in each layer of the input layer, hidden layer and output layer of the neural network is 4, 5 and 3. Therefore, the real number encoding of the firefly parameter individual is L = 3×5+5+5×3+3 = 38, and the weights and thresholds in the network are w ij 、w jk and θ j ,θ k , the firefly position vector can be expressed as:

[0104] x(t)=(w 11 …w ij , θ1…θ j , w 11 …w jk , θ1…θ k )

[0105] ② Construct the initial parameters of the BP neural network, including the initial threshold, number of training times, etc. Set the initial values ​​of the fireflies' emission, including the number n, the maximum number of iterations max, and the maximum attraction factor β0, and randomly initialize all individual positions.

[0106] ③ Calculate the objective function, that is, calculate the firefly fitness function. In the FA-BP neural network, the training accuracy is determined by the mean square error function E.

[0107] The individual fitness function f of fireflies is:

[0108]

[0109] ④ Determine the attraction and relative brightness between individual fireflies, the attraction β, and the distance between any two fireflies.

[0110] ⑤According to the fitness function, the position is moved to update the individual fitness function. In order to avoid the irregular movement of fireflies leading to the expansion of the effective distance of attraction, the lower limit of attraction β is added min , improve the search ability of FA.

[0111] x ad (t+1)=x ad (t)+(β-β min )(r ab [)x bd (t)-x ad (t)]+αε

[0112] ⑥When the number of iterations reaches the maximum or the error converges, the process ends; otherwise, return to the previous step.

[0113] ⑦ Map the global optimal position vector of the firefly to the BP neural network as the initial weight and threshold of the neural network after FA optimization. The BP neural network is trained until it reaches satisfaction, outputs the optimal parameters, and completes the training.

[0114] In order to verify the effectiveness of the FA-BP neural network algorithm, the BP neural network algorithm and the FA-BP neural network algorithm were used to fit the sample data respectively. Figure 2 shown.

[0115] Furthermore, in one embodiment of the present invention, when a deep learning network model is constructed based on mineral dissolution samples and the deep learning network model is used as an initial model for predicting mineral dissolution parameters, it also includes using a temporary dropout method (Dropout) behind a suitable hidden layer to partially discard the features learned by the previous hidden layer to prevent the model from overfitting during the training process; using an activation function in the hidden layer to perform a nonlinear transformation on the data, and using the output of the last hidden layer of the model through an activation function to obtain the final prediction result. And during the training process of the model, an optimization algorithm is used to adjust the network weights so that the predicted value of the mineral dissolution parameters obtained according to the deep learning network model converges with the loss function constructed by the actual mineral dissolution parameters, so as to train and obtain the mineral dissolution parameter prediction model.

[0116] In one embodiment of the present invention, the deep learning network model constructed includes an input layer, at least one hidden layer and an output layer. Exemplarily, the constructed deep learning network model can be as follows: Figure 3 As shown, the input layer includes 4 nodes (i.e., the dimension of the input mineral dissolution condition parameters is 4, for example, the input parameters can be ion concentration, mineral content, temperature and pressure conditions and mineral dissolution prediction period respectively), the output layer includes 3 nodes (i.e., the dimension of the predicted actual mineral dissolution parameters is 3, for example, the output is the ion concentration, mineral content and porosity after the mineral dissolution prediction period is reached), and the hidden layer is set to one layer and includes 5 neurons.

[0117] The temporary retreat method is to give a probability p, for example, p = 40%, then 40% of the neurons in the hidden layer are deleted, leaving only 60% of the neurons, such as Figure 4 As shown in the figure, for the deep learning network model, the essence is to assign the weight of these 40% neurons to 0, so the influence of neurons with weight assignment of 0 on the next layer (output layer) is 0. Then the remaining neurons are used to express the original hidden layer.

[0118] Through the above method, it is easy to adjust the complexity of the deep learning network model to avoid overfitting problems. At the same time, the reduction in the number of neurons in the hidden layer will not affect the entire deep learning network model.

[0119] Furthermore, in one embodiment of the present invention, before using the mineral dissolution sample to train the initial model for predicting mineral dissolution parameters in step S130, the method further includes:

[0120] The mineral dissolution samples are divided into training set and test set by using random sampling method or stratified sampling method;

[0121] The random sampling method is suitable for large data sets and for situations where the target values ​​are evenly distributed. For example, when training a binary classification model to handle classification tasks, when the training data set contains a large number of positive samples and a small number of negative samples (for example, negative samples only account for 10% of the data set), the label distribution of positive and negative examples is very uneven. If random sampling is used at this time, there may be an extreme situation where all positive samples are divided into the training set and all negative samples are divided into the test set. The effect of the binary classification model trained in this way will definitely not be very good. Therefore, at this time, it is preferred to use stratified sampling to divide the data set, so as to ensure that the training set contains both a certain proportion of positive samples and a certain proportion of negative samples.

[0122] In some feasible embodiments, 80% of the sample parameters can be used as a training set and 20% as a test set to prevent data snooping bias, avoid learning too much about the characteristics of the samples in the test set, and avoid selecting models that are helpful for the test set data.

[0123] The following formula is used to normalize the mineral dissolution condition parameters in the training set:

[0124]

[0125] Among them, x i is the mineral dissolution condition parameter of the i-th dimension in the training set, x i ′ is x i After normalization, the value of i ranges from 1 to n, where n is the dimension of the mineral dissolution condition parameter; μ i is the mean value of the mineral dissolution condition parameter in the i-th dimension, δ i is the standard deviation of the mineral dissolution condition parameters in the i-th dimension.

[0126] It should be noted that when other mineral dissolution condition parameters other than the mineral dissolution prediction period include more than one sub-dimension, the mean and standard deviation of the mineral dissolution condition parameter can be calculated based on the measured values ​​of the mineral dissolution condition parameter obtained at all observation times in multiple mineral dissolution samples for normalization. Normalization can overcome the problem of different dimensions between mineral dissolution parameters in different dimensions.

[0127] After using the normalized training set to train the initial model for predicting mineral dissolution parameters to obtain the trained mineral dissolution parameter prediction model, the test set is used to approximately estimate the generalization ability of the mineral dissolution parameter prediction model.

[0128] like Figure 5 FIG. 1 is a flow chart of a method for predicting mineral dissolution parameters provided by an embodiment of the present invention. The prediction method comprises:

[0129] S510: Collect mineral dissolution condition parameters;

[0130] The mineral dissolution condition parameters include at least one or a combination of ion concentration, mineral content, temperature and pressure conditions, acidity and alkalinity, and porosity; and at least the ion concentration, mineral content, temperature and pressure conditions, acidity and alkalinity, and porosity can be measurement values ​​obtained at a certain observation time, or can be measurement values ​​obtained at multiple observation times.

[0131] S520: Inputting the mineral dissolution condition parameters into the mineral dissolution parameter prediction model to obtain the mineral dissolution parameters predicted based on the mineral dissolution condition parameters.

[0132] For example, the mineral content (specifically the proportion of calcite and dolomite in the reservoir at multiple observation times) is input into a trained mineral dissolution parameter prediction model to predict the mineral dissolution parameters after a preset mineral dissolution prediction period. The mineral dissolution parameters can be the mineral content of the reservoir after the mineral dissolution prediction period, or can be parameters such as the porosity of the reservoir.

[0133] After the mineral dissolution parameters obtained according to the evolution of the mineral dissolution condition parameters are obtained, the mineral dissolution parameters can also be used to evaluate the reservoir quality.

[0134] like Figure 6 As shown, the present invention also provides a device for constructing a mineral dissolution parameter prediction model, comprising:

[0135] An acquisition module 61 is used to acquire a plurality of mineral dissolution samples, wherein the plurality of mineral dissolution samples each include a mineral dissolution condition parameter and an actual mineral dissolution parameter obtained according to the evolution of the mineral dissolution condition parameter;

[0136] A construction module 62 is used to construct an initial model for predicting mineral dissolution parameters according to the total dimension of the mineral dissolution condition parameters;

[0137] The training module 63 is used to train the initial model for predicting mineral dissolution parameters using the mineral dissolution samples until the error between the predicted values ​​of the mineral dissolution parameters obtained by the initial model for predicting mineral dissolution parameters and the actual mineral dissolution parameters is within a preset error range or the predicted values ​​of the mineral dissolution parameters reach a preset accuracy, thereby obtaining a trained mineral dissolution parameter prediction model.

[0138] like Figure 7 As shown, the present invention also provides a mineral dissolution parameter prediction device, comprising:

[0139] A collection module 71 is used to collect mineral dissolution condition parameters;

[0140] The prediction module 72 is used to input the mineral dissolution condition parameters into the mineral dissolution parameter prediction model to obtain the mineral dissolution parameters predicted according to the mineral dissolution condition parameters.

[0141] The beneficial effects obtained by the device provided by an embodiment of the present invention are consistent with the beneficial effects obtained by the above method, and will not be described in detail here.

[0142] like Figure 8 As shown, a computer device provided in an embodiment of the present invention, the computer device may be a mineral dissolution parameter prediction model training device provided in this specification to execute the mineral dissolution parameter prediction model training method provided in the present invention; the computer device may also be a mineral dissolution parameter prediction device to execute the mineral dissolution parameter prediction method provided in the present invention. The computer device 802 may include one or more processors 804, such as one or more central processing units (CPUs), each of which may implement one or more hardware threads. The computer device 802 may also include any memory 806 for storing any kind of information such as code, settings, data, etc. Non-limitingly, for example, the memory 806 may include any one or more combinations of the following: any type of RAM, any type of ROM, flash memory device, hard disk, optical disk, etc. More generally, any memory may use any technology to store information. Further, any memory may provide volatile or non-volatile retention of information. Further, any memory may represent a fixed or removable component of the computer device 802. In one embodiment, when the processor 804 executes the associated instructions stored in any memory or combination of memories, the computer device 802 can perform any operation of the associated instructions. The computer device 802 also includes one or more drive mechanisms 808 for interacting with any memory, such as a hard disk drive mechanism, an optical disk drive mechanism, etc.

[0143] The computer device 802 may also include an input / output module 810 (I / O) for receiving various inputs (via input devices 812) and for providing various outputs (via output devices 814). A specific output mechanism may include a presentation device 816 and an associated graphical user interface (GUI) 818. In other embodiments, the input / output module 810 (I / O), the input device 812, and the output device 814 may not be included, and the computer device 802 may be used as a computer device in a network. The computer device 802 may also include one or more network interfaces 820 for exchanging data with other devices via one or more communication links 822. One or more communication buses 824 couple the components described above together.

[0144] The communication link 822 may be implemented in any manner, for example, through a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication link 822 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc. governed by any protocol or combination of protocols.

[0145] Corresponding to Figure 1 and Figure 5 The method shown in the embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are executed.

[0146] The embodiment of the present invention also provides a computer-readable instruction, wherein when a processor executes the instruction, the program therein causes the processor to execute the following Figure 1 and Figure 5 The method shown in .

[0147] The embodiment of the present invention further provides a computer program product, comprising at least one instruction or at least one program segment, wherein the at least one instruction or the at least one program segment is loaded and executed by a processor to implement the following Figure 1 and Figure 5 The method shown in .

[0148] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0149] It should also be understood that in the embodiments of the present invention, the term "and / or" is only a description of the association relationship of the associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in the present invention generally indicates that the associated objects before and after are in an "or" relationship.

[0150] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0151] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0152] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or it can be an electrical, mechanical or other form of connection.

[0153] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present invention.

[0154] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0155] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0156] 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 are included in the protection scope of the present invention.

Claims

1. A method for constructing a mineral dissolution parameter prediction model, characterized in that: The steps include: Acquire a plurality of mineral dissolution parameter action samples, wherein the plurality of mineral dissolution parameter action samples all include mineral dissolution condition parameters and actual mineral dissolution parameters obtained according to the evolution of the mineral dissolution condition parameters; Constructing an initial model for predicting mineral dissolution parameters according to the mineral dissolution sample and the total dimension of the mineral dissolution condition parameters; The mineral dissolution parameter prediction initial model is trained using the mineral dissolution parameter action sample until the error between the mineral dissolution parameter prediction value obtained by the mineral dissolution parameter prediction initial model and the actual mineral dissolution parameter is within a preset error range or the mineral dissolution parameter prediction value reaches a preset accuracy rate, thereby obtaining a mineral dissolution parameter prediction model.

2. The method for constructing a mineral dissolution parameter prediction model according to claim 1, characterized in that: Before using the mineral dissolution parameter action sample to train the initial model for predicting the mineral dissolution parameters, it also includes: performing feature selection on the mineral dissolution condition parameters, and removing the mineral dissolution condition parameters whose influence coefficient on the actual mineral dissolution parameters is less than a preset value.

3. The method for constructing a mineral dissolution parameter prediction model according to claim 1, characterized in that: According to the mineral dissolution sample and the total dimension of the mineral dissolution condition parameters, an initial model for predicting mineral dissolution parameters is constructed, which specifically includes the following steps: When the total dimension of the mineral dissolution condition parameters is less than a preset dimension threshold, a machine learning model is constructed according to the mineral dissolution sample, and the machine learning model is used as an initial model for predicting mineral dissolution parameters; the algorithm used by the machine learning model includes at least one of a decision tree algorithm, a naive Bayes algorithm, a support vector machine algorithm, a random forest algorithm, an artificial neural network algorithm, a Boosting and Bagging algorithm, an association rule algorithm, and an expectation maximization algorithm; When the total dimension of the mineral dissolution condition parameters is greater than or equal to a preset dimension threshold, a deep learning network model is constructed according to the mineral dissolution sample, and the deep learning network model is used as an initial model for predicting mineral dissolution parameters.

4. The method for constructing a mineral dissolution parameter prediction model according to claim 1, characterized in that: After obtaining a plurality of mineral dissolution parameter action samples, and before using the mineral dissolution parameter action samples to train the mineral dissolution parameter prediction initial model, the method further includes: The mineral dissolution samples are divided into training sets and test sets according to a preset ratio by using a random sampling method or a stratified sampling method; The following formula is used to normalize the mineral dissolution condition parameters in the training set: Among them, x i is the mineral dissolution condition parameter of the i-th dimension in the training set, x i ′ is x i After normalization, the value of i ranges from 1 to n, where n is the dimension of the mineral dissolution condition parameter; μ i is the mean value of the mineral dissolution condition parameter in the i-th dimension, δ i is the standard deviation of the mineral dissolution condition parameters in the i-th dimension.

5. The method for constructing a mineral dissolution parameter prediction model according to any one of claims 1 to 4, characterized in that: The dissolution condition parameters include a mineral dissolution prediction period, and the dissolution condition parameters also include one or a combination of several of ion concentration, mineral content, temperature and pressure conditions, acidity and alkalinity, and porosity.

6. The method for constructing a mineral dissolution parameter prediction model according to claim 5, characterized in that: The actual dissolution parameters include at least one or a combination of ion concentration, mineral content, temperature and pressure conditions, acidity and alkalinity, and porosity after the evolution time reaches the predicted period of mineral dissolution.

7. The method for constructing a mineral dissolution parameter prediction model according to claim 6, characterized in that: The dissolution condition parameter values ​​in the mineral dissolution second parameter set in the mineral dissolution condition parameters are measured values ​​obtained at a certain observation time, or are measured values ​​obtained at multiple observation times.

8. A method for predicting mineral dissolution parameters, characterized in that: The steps include: Collect mineral dissolution condition parameters; The collected mineral dissolution condition parameters are input into the mineral dissolution parameter prediction model obtained by the method for constructing a mineral dissolution parameter prediction model as described in any one of claims 1 to 7 to obtain predicted mineral dissolution parameters.

9. A device for constructing a mineral dissolution parameter prediction model, characterized in that: include: An acquisition module, used for acquiring a plurality of mineral dissolution parameter action samples, wherein the plurality of mineral dissolution parameter action samples all include mineral dissolution condition parameters and actual mineral dissolution parameters obtained according to the evolution of the mineral dissolution condition parameters; A construction module, for constructing an initial model for predicting mineral dissolution parameters according to the mineral dissolution sample and the total dimension of the mineral dissolution condition parameters; A training module is used to train the initial model for predicting mineral dissolution parameters using the mineral dissolution parameter action sample, until the error between the predicted value of the mineral dissolution parameter obtained by the initial model for predicting mineral dissolution parameters and the actual mineral dissolution parameter is within a preset error range or the predicted value of the mineral dissolution parameter reaches a preset accuracy, thereby obtaining a mineral dissolution parameter prediction model.

10. A mineral dissolution parameter prediction device, characterized in that: include: Collection module, used to collect mineral dissolution condition parameters; A prediction module is used to input the mineral dissolution condition parameters into the mineral dissolution parameter prediction model obtained by the mineral dissolution parameter prediction model construction device according to claim 9 to obtain the mineral dissolution parameters predicted based on the mineral dissolution condition parameters.

11. A computer device, characterized in that: It includes a processor, which can execute the mineral dissolution parameter prediction model construction method described in any one of claims 1 to 7 when running.

Images