Method and device for establishing and predicting dolomitic parameter data prediction model

Abstract

The invention relates to the technical field of reservoir parameter prediction, and provides a dolomitization parameter data prediction model establishment method, a dolomitization parameter data prediction method and a dolomitization parameter data prediction device. The dolomitization parameter data prediction model establishment method comprises the steps of obtaining a dolomitization parameter data set corresponding to a dolomitization sample; screening the dolomitization parameter data set to obtain a screening result; constructing a training sample according to the screening result; utilizing a hyper-parameter optimization algorithm to optimize hyper-parameters in the gradient boosting decision tree model; and training the optimized gradient boosting decision tree model by using the training sample to obtain a dolomitization parameter data prediction model. According to the embodiment of the invention, the evaluation precision of dolomitization parameters can be improved.

Description

Technical Field

[0001] The present specification relates to the technical field of reservoir parameter prediction, and in particular to the establishment of a dolomitization parameter data prediction model, a prediction method and a device. Background Art

[0002] Dolomitization refers to the process of converting loose sediments into sedimentary rocks under the influence of certain pressure and temperature. The mechanism and quality evaluation of reservoir dolomitization are key issues facing oil and gas exploration and development today, and studying the evolution process of reservoir dolomitization is crucial for the mechanism and quality evaluation of reservoir dolomitization. Based on the history of burial diagenesis, petrography and geochemical characteristics, the existing technology can achieve quantitative evaluation of dolomitization parameters such as mineral content, ion concentration, reservoir porosity, and permeability in dolomitization to a certain extent through numerical simulation methods. However, due to the fact that the reservoir dolomitization time is as long as one million years and there are many influencing factors, the numerical simulation of dolomitization parameters has the problems of large amount of calculation, high uncertainty and strong subjectivity, which affects the accuracy of reservoir evaluation and has limited effect on oil and gas exploration. Therefore, there is an urgent need for a method for establishing a dolomitization parameter data prediction model to establish a dolomitization parameter data prediction model, thereby improving the evaluation accuracy of dolomitization parameters. Summary of the invention

[0003] Considering that the dolomitization time of the reservoir is as long as one million years and there are many influencing factors, the numerical simulation of dolomitization parameters has the problems of large computational complexity, high uncertainty and strong subjectivity, which affects the accuracy of reservoir evaluation and has limited effect on oil and gas exploration. This scheme is proposed to overcome the above problems or at least partially solve the above problems.

[0004] On the one hand, the purpose of some embodiments of this specification is to provide a method for establishing a dolomitization parameter data prediction model, the method comprising:

[0005] Obtain dolomitization parameter data sets corresponding to dolomitization samples;

[0006] Screening the dolomitization parameter data set to obtain a screening result;

[0007] Constructing a training sample according to the screening results;

[0008] Use hyperparameter optimization algorithms to optimize hyperparameters in the gradient boosting decision tree model;

[0009] The training samples are used to train the optimized gradient boosting decision tree model to obtain a dolomitization parameter data prediction model.

[0010] Furthermore, the dolomitization parameter data set is screened, and before obtaining the screening result, the method further comprises:

[0011] According to the attribute characteristics of the dolomitization parameter data set, the dolomitization parameter data set is classified to obtain several classified data sets; different classified data sets include corresponding several attribute characteristic data sets.

[0012] Furthermore, the dolomitization parameter data set is screened to obtain screening results, including:

[0013] Initialize the weights of all attribute feature data sets;

[0014] From the dolomitization parameter data set, a property feature data set is selected by using a random algorithm as a central data set;

[0015] From the classification data set where the central data set is located, a property feature data set is selected using a random algorithm as the first nearest neighbor data set;

[0016] From the classification data sets other than the classification data set where the central data set is located, a property feature data set is selected by using a random algorithm as the second nearest neighbor data set;

[0017] Calculate the first distance and the second distance between the central data set and the first nearest neighbor data set and the second nearest neighbor data set respectively;

[0018] comparing the first distance and the second distance;

[0019] If the first distance is not greater than the second distance, the weight of the central data set is updated using the weight gain coefficient;

[0020] If the first distance is greater than the second distance, the weight of the central data set is updated using the weight decay coefficient;

[0021] Repeat the steps of randomly selecting a central data set, a first nearest neighbor data set, and a second nearest neighbor data set, and updating the weight of the central data set according to the first distance and the second distance between the central data set and the first nearest neighbor data set and the second nearest neighbor data set, until a preset convergence condition is reached, thereby updating the weights of all attribute feature data sets;

[0022] The attribute feature data set whose weight is less than a preset threshold is removed from the dolomitization parameter data set to obtain the screening result.

[0023] Furthermore, the base learner of the gradient boosting decision tree model is a symmetric decision tree.

[0024] Furthermore, the hyperparameter optimization algorithm is used to optimize the hyperparameters in the gradient boosting decision tree model, including:

[0025] Randomly initialize to obtain hyperparameter data to establish a corresponding optimization space; the hyperparameter data includes at least one or more of the number of iterations, learning rate, and maximum depth;

[0026] Establish the objective function for hyperparameter optimization;

[0027] In the optimization space, the objective function is subjected to Gaussian process processing to obtain a marginal likelihood distribution;

[0028] Calculating the optimal solution of the current hyperparameters according to the marginal likelihood distribution;

[0029] According to the current optimal solution of the hyperparameters, the objective function is optimized and updated using a preset acquisition function to obtain the optimal hyperparameters;

[0030] The gradient boosting decision tree model is optimized according to the optimal hyperparameters.

[0031] Furthermore, the hyperparameter data is randomly initialized to establish the corresponding optimization space, including:

[0032] In the preset mapping space, the distribution space of hyperparameters is randomly initialized using chaotic mapping;

[0033] According to the preset number of data required, the hyperparameter data are selected from the distribution space, and each type of hyperparameter data conforms to the uniform distribution, so as to establish an optimization space according to the hyperparameter data.

[0034] Furthermore, the optimized gradient boosting decision tree model is trained using the training samples to obtain a dolomitization parameter data prediction model, including:

[0035] Initializing a learner; the learner includes at least one decision tree;

[0036] Using the training samples to train the learner to obtain a prediction result;

[0037] According to the real labels and prediction results in the training samples, the loss is calculated using a preset loss function;

[0038] Calculate the negative gradient based on the loss and the prediction result;

[0039] Performing weighted random sampling on the training samples using the negative gradient to obtain sampled training samples;

[0040] Using the sampled training samples to train a new decision tree;

[0041] Scaling the trained new decision tree according to a preset learning rate, and adding the scaling result to the learner to update the learner;

[0042] Repeat the steps of training a new decision tree using the sampled training samples and updating the learner according to the trained new decision tree until the number of iterations reaches a preset threshold.

[0043] On the other hand, some embodiments of this specification also provide a dolomitization parameter data prediction model establishment device, the device comprising:

[0044] An acquisition module, used to acquire a dolomitization parameter data set corresponding to a dolomitization sample;

[0045] A screening module, used for screening the dolomitization parameter data set to obtain a screening result;

[0046] A construction module, used to construct a training sample according to the screening result;

[0047] The optimization module is used to optimize the hyperparameters in the gradient boosting decision tree model using the hyperparameter optimization algorithm;

[0048] The training module is used to train the optimized gradient boosting decision tree model using the training samples to obtain a dolomitization parameter data prediction model.

[0049] Based on the same inventive concept, some embodiments of this specification also provide a dolomitization parameter data prediction method, the method comprising:

[0050] receiving a dolomitization parameter data set to be processed;

[0051] The dolomitization parameter data set to be processed is input into a dolomitization parameter data prediction model trained by the method described in any of the aforementioned embodiments to predict corresponding dolomitization parameter prediction data.

[0052] Based on the same inventive concept, on the other hand, some embodiments of this specification further provide a dolomitization parameter data prediction device, the device comprising:

[0053] A receiving module, used for receiving a dolomitization parameter data set to be processed;

[0054] The prediction module is used to input the dolomitization parameter data set to be processed into the dolomitization parameter data prediction model trained by the method described in any of the above embodiments to predict the corresponding dolomitization parameter prediction data.

[0055] On the other hand, some embodiments of the present specification further provide a computer device, including a memory, a processor, and a computer program stored in the memory, wherein when the computer program is executed by the processor, the instructions of the above method are executed.

[0056] On the other hand, some embodiments of the present specification further provide a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor of a computer device, the instructions of the above method are executed.

[0057] On the other hand, some embodiments of the present specification further provide a computer program product, wherein the computer program product comprises a computer program, and when the computer program is executed by a processor of a computer device, the instructions of the above method are executed.

[0058] One or more technical solutions provided by some embodiments of this specification have at least the following technical effects:

[0059] The embodiments of the present specification first obtain a dolomite parameter data set corresponding to a dolomite sample. Since the dolomite parameter data set involves a large amount of data and multiple data dimensions, the dolomite parameter data set is screened to achieve dimensionality reduction and remove redundant data. A training sample is then constructed based on the screening results, and the hyperparameters in the gradient boosting decision tree model are optimized through a hyperparameter optimization algorithm to improve the model performance of the gradient boosting decision tree model. The optimized gradient boosting decision tree model is trained using the training samples, thereby quickly obtaining a dolomite parameter data prediction model to achieve accurate and objective prediction of dolomite parameter data.

[0060] The above description is only an overview of the technical solutions of some embodiments of this specification. In order to more clearly understand the technical means of some embodiments of this specification, they can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of some embodiments of this specification more obvious and easy to understand, the specific implementation methods of some embodiments of this specification are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate some embodiments of this specification or technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or prior art descriptions. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. In the drawings:

[0062] Figure 1 A schematic diagram of an implementation system of a dolomitization parameter data prediction method in some embodiments of this specification is shown;

[0063] Figure 2 A flowchart of a method for establishing a dolomitization parameter data prediction model in some embodiments of this specification is shown;

[0064] Figure 3A schematic diagram of the steps of screening a dolomitization parameter data set in some embodiments of this specification;

[0065] Figure 4 A schematic diagram of the steps for optimizing hyperparameters in a gradient boosting decision tree model in some embodiments of this specification;

[0066] Figure 5 A schematic diagram of the steps of establishing an optimization space in some embodiments of this specification;

[0067] Figure 6 A schematic diagram of the steps of using training samples to train an optimized gradient boosting decision tree model in some embodiments of this specification;

[0068] Figure 7 A flowchart of a dolomitization parameter data prediction method in some embodiments of this specification is shown;

[0069] Figure 8 This is a schematic diagram of the structure of a device for establishing a dolomitization parameter data prediction model in some embodiments of this specification;

[0070] Fig. 9 This is a schematic diagram of the structure of a dolomitization parameter data prediction device in some embodiments of this specification;

[0071] Fig.10 This is a schematic diagram of the computer device structure provided in some embodiments of this specification.

[0072] [Description of Reference Numerals]

[0073] 101. Terminal;

[0074] 102. Server;

[0075] 801, obtaining module;

[0076] 802, screening module;

[0077] 803, building blocks;

[0078] 804. Optimization module;

[0079] 805. Training module;

[0080] 901, receiving module;

[0081] 902. prediction module;

[0082] 1002. Computer equipment;

[0083] 1004. Processor;

[0084] 1006. Memory;

[0085] 1008. Driving mechanism;

[0086] 1010, input / output interface;

[0087] 1012. Input device;

[0088] 1014. Output device;

[0089] 1016. Presentation equipment;

[0090] 1018. Graphical user interface;

[0091] 1020. Network interface;

[0092] 1022. Communication link;

[0093] 1024. Communication bus. DETAILED DESCRIPTION

[0094] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in some embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on some embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0095] It should be noted that the terms "first", "second", etc. in the specification and claims of this article and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of this article described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, device, product or equipment that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0096] It should be noted that the acquisition, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of relevant laws and regulations.

[0097] It should be noted that in the embodiments of this specification, certain software, components, models and other existing solutions in the industry may be mentioned, and they should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of this application, but it does not mean that the applicant has or will necessarily use the solution.

[0098] like Figure 1 The diagram is a schematic diagram of an implementation system of a method for establishing a dolomitization parameter data prediction model according to an embodiment of the present invention, which may include: a terminal 101 and a server 102, wherein the terminal 101 and the server 102 communicate with each other through a network, and the network may include a local area network (LAN), a wide area network (WAN), the Internet, or a combination thereof, and is connected to a website, a user device (such as a computing device), and a back-end system. A staff member may send a dolomitization parameter data prediction model establishment request to the server 102 through the terminal 101, and after receiving the dolomitization parameter data prediction model establishment request, the server 102 calls the data in the database for calculation and processing to obtain a modeling result, and sends the modeling result to the terminal 101, so that the staff member processes the business according to the modeling result.

[0099] In the embodiments of this specification, the server 102 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN, Content Delivery Network), and big data and artificial intelligence platforms.

[0100] In an optional embodiment, the terminal 101 may include but is not limited to electronic devices such as self-service terminal devices, desktop computers, tablet computers, and laptop computers. Optionally, the operating system running on the electronic device may include but is not limited to Android system, IOS system, Linux, Windows, etc. Of course, the terminal 101 is not limited to the above-mentioned electronic devices with a certain entity, and it can also be software running in the above-mentioned electronic devices.

[0101] In addition, it should be noted that Figure 1 What is shown is only an application environment provided by the present disclosure. In actual application, multiple terminals 101 may be included, and this specification does not limit this.

[0102] Figure 2It is a flowchart of a method for establishing a dolomitization parameter data prediction model provided by an embodiment of the present invention. This specification provides method operation steps as described in the embodiment or flowchart, but may include more or fewer operation steps 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 actual system or device product is executed, it can be executed in the order or in parallel according to the method shown in the embodiment or the accompanying drawings. Specifically, Figure 2 As shown, applied to the above-mentioned server side, the method may include:

[0103] S201: Obtain a dolomitization parameter data set corresponding to a dolomitization sample;

[0104] Dolomitization refers to the process of converting loose sediments into sedimentary rocks under certain pressure and temperature. The reservoir dolomitization time is as long as millions of years, and there are hundreds of influencing factors. The existing numerical simulation calculation methods of dolomitization have the problems of large calculation amount, strong uncertainty and large error.

[0105] The attribute characteristics of the dolomitization parameter data set may include one or more of ion concentration, mineral content, temperature and pressure conditions, acidity and alkalinity, porosity, hydrodynamic conditions, longitude and latitude, burial depth, etc., which are not limited in this article.

[0106] In some embodiments, the dolomitization parameter data set often includes dolomitization parameter data in a real geological environment. In other embodiments, the dolomitization parameter data set may also include simulation data obtained through simulation and consistent with the real geological environment, which is not limited herein. Furthermore, in some embodiments, in order to improve the data richness of the dolomitization parameter data set, the dolomitization parameter data set may also include dolomitization parameter data at multiple historical moments.

[0107] In addition, when using the dolomitization parameter data set to predict dolomitization parameter data, it is also necessary to predetermine the prediction time corresponding to the dolomitization parameter data, so as to establish a matching dolomitization parameter data prediction model according to the prediction time.

[0108] S202: screening the dolomitization parameter data set to obtain screening results;

[0109] The more attribute features a dolomitization parameter dataset has, the larger its data dimension is. In order to ensure the richness of the dolomitization parameter dataset, a dolomitization parameter dataset with a larger data dimension needs to be obtained in the early stage. However, too large a data dimension will lead to high subsequent model training costs and affect the accuracy of model prediction. Therefore, in some embodiments, it is also necessary to screen the dolomitization parameter dataset to remove redundant data. Specifically, the dolomitization parameter dataset is screened, and before obtaining the screening results, it further includes:

[0110] According to the attribute characteristics of the dolomitization parameter data set, the dolomitization parameter data set is classified to obtain several classified data sets; different classified data sets include corresponding several attribute characteristic data sets.

[0111] It can be understood that, in some embodiments, at a preset prediction time, the dolomitization parameter data set can be classified according to the attribute characteristics of the dolomitization parameter data set, and each classified data set obtained by classification includes several corresponding attribute characteristic data sets. For example, for classification data set A, it includes two attribute characteristic data sets a1 and a2, and for classification data set B, it includes two attribute characteristic data sets b1, b2 and b3.

[0112] Specifically, since the global similarity probability between two attribute feature data sets in the same classification data set is often higher than the global similarity probability between two attribute feature data sets in different classification data sets, classifying the dolomite parameter data set by attribute features can facilitate the subsequent screening of the dolomite parameter data set by comparing the distances between attribute feature data sets based on different classification data sets, thereby reducing the computational cost.

[0113] Furthermore, in some embodiments, the dolomitization parameter data set is screened, and before the screening result is obtained, the dolomitization parameter data set may be preprocessed, and the preprocessing process may include normalization processing and the like.

[0114] See attached Figure 3 In some embodiments, screening the dolomitization parameter data set to obtain screening results may include:

[0115] S301: Initialize the weights of all attribute feature data sets;

[0116] S302: Selecting an attribute feature data set from the dolomitization parameter data set using a random algorithm as a central data set;

[0117] S303: Selecting an attribute feature data set from the classification data set where the central data set is located by using a random algorithm as the first nearest neighbor data set;

[0118] S304: Selecting an attribute feature dataset from other classification datasets except the classification dataset where the central dataset is located by using a random algorithm as the second nearest neighbor dataset;

[0119] S305: Calculate the first distance and the second distance between the central data set and the first nearest neighbor data set and the second nearest neighbor data set respectively;

[0120] S306: Compare the first distance and the second distance;

[0121] S307: If the first distance is not greater than the second distance, updating the weight of the central data set using the weight gain coefficient;

[0122] S308: If the first distance is greater than the second distance, updating the weight of the central data set using the weight decay coefficient;

[0123] S309: Repeat the steps of randomly selecting a central data set, a first neighbor data set, and a second neighbor data set, and updating the weight of the central data set according to the first distance and the second distance between the central data set and the first neighbor data set and the second neighbor data set, until a preset convergence condition is reached, thereby updating the weights of all attribute feature data sets;

[0124] S3010: Removing the attribute feature data set whose weight is less than a preset threshold from the dolomitization parameter data set to obtain the screening result.

[0125] It can be understood that in some embodiments, the weights of all attribute feature data sets can be initialized by a random algorithm, or they can be uniformly given the same initial weights. The random algorithm can be a Monte Carlo algorithm, a Las Vegas algorithm, etc., which is not limited in this article.

[0126] Afterwards, in each iteration process, an attribute feature dataset is randomly selected from the dolomitization parameter dataset as the central dataset, an attribute feature dataset is randomly selected from the classification dataset where the central dataset is located as the first nearest neighbor dataset, and an attribute feature dataset is randomly selected from other classification datasets except the classification dataset where the central dataset is located as the second nearest neighbor dataset. The first distance between the central dataset and the first nearest neighbor dataset, as well as the second distance between the central dataset and the second nearest neighbor dataset are calculated respectively. The ability of the central dataset to distinguish the first nearest neighbor dataset from the second nearest neighbor dataset is judged by comparing the first distance and the second distance.

[0127] Specifically, since the first neighbor data set and the central data set are in the same classification data set, and the second neighbor data set and the central data set are in different classification data sets, when the first distance is not greater than the second distance, it means that the central data set is good at distinguishing the first neighbor data set from the second neighbor data set. At this time, the weight of the central data set is used with the weight gain coefficient. Otherwise, it means that the central data set is not good at distinguishing the first neighbor data set that is in the same classification data set as the central data set and the second neighbor data set that is in a different classification data set from the central data set. At this time, the weight attenuation coefficient is used to reduce the weight of the central data set.

[0128] Afterwards, the steps of randomly selecting the central data set, the first nearest neighbor data set, and the second nearest neighbor data set are repeated, and the weight of the central data set is updated according to the first distance and the second distance between the central data set and the first nearest neighbor data set and the second nearest neighbor data set, until a preset convergence condition is reached, and the preset convergence condition can be a preset number of iterations, etc., so as to obtain the updated weights of all attribute feature data sets. Among them, the attribute feature data sets with weights less than the preset threshold are redundant data sets with weak distinguishing ability. Therefore, by removing the attribute feature data sets with weights less than the preset threshold, the screened dolomitization parameter data set can be quickly obtained.

[0129] S203: constructing training samples according to the screening results;

[0130] Training samples can be constructed based on the screened dolomitization parameter data set. The training samples include input data and labeled target output data. The training samples can be used to construct training sets, validation sets, and test sets for subsequent training of gradient boosting decision tree models.

[0131] S204: Optimizing hyperparameters in the gradient boosting decision tree model using a hyperparameter optimization algorithm;

[0132] In some embodiments, the base learner of the gradient boosting decision tree model is a symmetric decision tree. The symmetric decision tree is more balanced than the conventional decision tree. In the symmetric decision tree, the same segmentation criterion is used on the entire layer of the tree, which can effectively prevent overfitting and greatly increase the reliability of the model.

[0133] Further, after constructing the training samples, refer to the attached Figure 4 In some embodiments, optimizing hyperparameters in a gradient boosting decision tree model using a hyperparameter optimization algorithm may include:

[0134] S401: Randomly initialize to obtain hyperparameter data to establish a corresponding optimization space;

[0135] Specifically, the hyperparameter data includes at least one or more of the number of iterations, learning rate, and maximum depth. Figure 5 In some embodiments, randomly initializing to obtain hyperparameter data to establish a corresponding optimization space may include:

[0136] S501: In a preset mapping space, randomly initialize the distribution space of hyperparameters using chaotic mapping;

[0137] S502: According to a preset number of data acquisition requirements, the hyperparameter data is selected from the distribution space, and each type of hyperparameter data conforms to a uniform distribution, so as to establish an optimization space according to the hyperparameter data.

[0138] It can be understood that, in some embodiments, the range of the preset mapping space is larger than the range of the commonly used empirical space of the hyperparameters (i.e., the mapping subspace), and the mapping space provides a wider data acquisition range compared to the mapping subspace. In order to improve the randomness of subsequent hyperparameter data, the distribution space of the hyperparameters is randomly initialized using chaotic mapping within the preset mapping space, and the distribution space always includes the above-mentioned mapping subspace to avoid the situation where the distribution space is too small, thereby ensuring the effectiveness of the subsequent data acquisition process.

[0139] Afterwards, according to the preset number of data required (for example, the number of data must be greater than 100), hyperparameter data is selected from the distribution space. At the same time, each hyperparameter data conforms to the uniform distribution, so as to establish an optimization space based on the hyperparameter data. The hyperparameter optimization algorithm will select a suitable set of hyperparameter data in the optimization space to optimize the hyperparameter settings in the gradient boosting decision tree model.

[0140] S402: Establishing an objective function for hyperparameter optimization;

[0141] S403: In the optimization space, performing Gaussian process processing on the objective function to obtain marginal likelihood distribution;

[0142] S404: Obtaining the optimal solution of the current hyperparameters according to the marginal likelihood distribution calculation;

[0143] S405: According to the current optimal solution of the hyperparameters, the objective function is optimized and updated using a preset acquisition function to obtain the optimal hyperparameters;

[0144] S406: Optimizing the gradient boosting decision tree model according to the optimal hyperparameters.

[0145] It can be understood that in some embodiments, the objective function is a generalized function for hyperparameter optimization, and only the input and output need to be specified without knowing its internal structure and mathematical properties. Specifically, assuming that the training set is M, η i is a sample in the training set M, each sample represents an observation result, and the set of unknown objective functions is {g(η1),g(η2),…,g(ηn-1 )}, n represents the number of samples, the hyperparameter is θ, and the hyperparameter represents an assumption. When there is observation noise ε and the noise is assumed to satisfy an independent and identically distributed Gaussian distribution P(ε) = (μ,σ 2 ), the objective function obeys the Gaussian distribution, and the Gaussian distribution can be understood as a priori model, σ 2 is the predicted variance of the prior model, μ is the predicted mean of the prior model, and the marginal likelihood distribution between the sample and the hyperparameter can be obtained:

[0146]

[0147] Among them, P(η i |M,θ) is the marginal likelihood distribution function, θ best It is the optimal solution for the current hyperparameters and can be obtained by maximizing the marginal likelihood distribution function using the maximum likelihood estimation algorithm.

[0148] It should be noted that in statistics, the a priori model refers to the assumptions made about the distribution of parameters or variables before the data is observed. It is based on experience or previous knowledge, not on specific data. In contrast, the posterior model is the inference made about the distribution of parameters or variables after the data is observed.

[0149] Furthermore, after obtaining the optimal solution for the current hyperparameters, since the optimal solution for the current hyperparameters is still a judgment result of uncertainty, and the posterior distribution corresponding to the objective function does not fit the actual distribution at this time, the acquisition function needs to update the hyperparameters according to the current optimal solution for the hyperparameters and μ and σ given by the prior model, so as to achieve the optimal update of the objective function, thereby continuously updating the hyperparameters through the acquisition function until the posterior distribution basically fits the true distribution, and the final optimal hyperparameters are obtained to optimize the gradient boosting decision tree model.

[0150] In some embodiments, the posterior distribution can be calculated using the following formula:

[0151]

[0152] Among them, p(f|D 1:t ) indicates the observed data D 1:t The posterior distribution of the modified parameter f, f represents the parameter in the objective function, p(D 1:t ) represents the marginal likelihood distribution of parameter f, p(f) represents the prior distribution of parameter f, p(D 1:t |f) represents the likelihood distribution of observed data under parameter f, D 1:t Indicates that the size of the observed set is t.

[0153] In some embodiments, the acquisition function can be constructed using the following formula:

[0154] α(θ|μ,σ)=E[max(0,g(θ)-g(θ best ))]

[0155] Among them, α(θ|μ,σ) is the acquisition function, μ is the predicted mean of the prior model, σ is the predicted standard deviation of the prior model, E is the current optimal expectation, g(·) is the objective function, θ is the hyperparameter, and θ best is the optimal solution for the current hyperparameters.

[0156] S205: Using the training samples to train the optimized gradient boosting decision tree model to obtain a dolomitization parameter data prediction model.

[0157] See attached Figure 6 In some embodiments, using the training samples to train the optimized gradient boosting decision tree model to obtain a dolomitization parameter data prediction model may include:

[0158] S601: Initialize a learner; the learner includes at least one decision tree;

[0159] S602: Train the learner using the training sample to obtain a prediction result;

[0160] S603: Calculate the loss using a preset loss function according to the real labels and prediction results in the training samples;

[0161] S604: Calculate a negative gradient according to the loss and the prediction result;

[0162] S605: Performing weighted random sampling on the training samples using the negative gradient to obtain sampled training samples;

[0163] S606: Train a new decision tree using the sampled training samples;

[0164] S607: Scaling the trained new decision tree according to a preset learning rate, and adding the scaling result to the learner to update the learner;

[0165] S608: Repeat the steps of training a new decision tree using the sampled training samples and updating the learner according to the trained new decision tree until the number of iterations reaches a preset threshold.

[0166] It can be understood that, in some embodiments, the decision tree in the learner is a symmetric decision tree, and the process of the initialized learner obtaining the prediction result can be expressed as:

[0167]

[0168] Among them, F0(x) is the output of the learner, argmin is the value of γ when the latter formula reaches the minimum value, n is the number of samples, L is the loss function, and y i is the true label of the i-th sample, and γ is the model predicted label.

[0169] The negative gradient is calculated using the following formula:

[0170]

[0171] Among them, g im is the negative gradient, m is the number of iterations, i is the sample number, y i is the true label of the i-th sample, F m-1 (x i ) is the prediction result of the mth i-th sample, x i is the input data of the i-th sample, and L is the loss function.

[0172] After calculating the negative gradient, based on the current training sample, im Perform weighted random sampling for the weights to obtain sampled training samples, and then train a new decision tree based on the sampled training samples to update the learner, and recalculate the corresponding loss and negative gradient based on the new decision tree, and resample to obtain new sampled training samples, so as to use the new sampled training samples to train another new decision tree, and repeat the steps of training a new decision tree with the sampled training samples, and updating the learner according to the trained new decision tree, until the number of iterations reaches the preset threshold, and finally output the trained gradient boosting decision tree model:

[0173] F m (x) = F m-1 (x)+ηT m (x)

[0174] Among them, F m (X) is the trained gradient boosting decision tree model, m is the number of iterations, and F m-1 (x) is the gradient boosting decision tree model trained in the previous iteration, T m (x) is the decision tree newly added in the current iteration, and η is the learning rate, which is used to control T m The contribution size of (x).

[0175] Some embodiments of the present specification first obtain a dolomite parameter data set corresponding to a dolomite sample. Since the dolomite parameter data set involves a large amount of data and multiple data dimensions, the dolomite parameter data set is screened to achieve dimensionality reduction and remove redundant data. A training sample is then constructed based on the screening results, and the hyperparameters in the gradient boosting decision tree model are optimized through a hyperparameter optimization algorithm to improve the model performance of the gradient boosting decision tree model. The optimized gradient boosting decision tree model is trained using the training samples, thereby quickly obtaining a dolomite parameter data prediction model to achieve accurate and objective prediction of dolomite parameter data.

[0176] Based on the same inventive concept, some embodiments of this specification also provide a dolomitization parameter data prediction method, referring to Figure 7 As shown, in some embodiments, the method may include:

[0177] S701: receiving a dolomitization parameter data set to be processed;

[0178] S702: Inputting the dolomitization parameter data set to be processed into a dolomitization parameter data prediction model trained by the method described in any of the above embodiments to predict corresponding dolomitization parameter prediction data.

[0179] It should be noted that, although the operations of the method of the present invention are described in a specific order in the above embodiments and the accompanying drawings, this does not require or imply that the operations must be performed in the specific order, or that all the operations shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0180] Corresponding to the above-mentioned method for establishing a dolomitization parameter data prediction model, some embodiments of this specification also provide a dolomitization parameter data prediction model establishment device, referring to Figure 8 As shown, in some embodiments, the apparatus may include:

[0181] An acquisition module 801 is used to acquire a dolomitization parameter data set corresponding to a dolomitization sample;

[0182] A screening module 802 is used to screen the dolomitization parameter data set to obtain a screening result;

[0183] A construction module 803 is used to construct a training sample according to the screening result;

[0184] An optimization module 804 is used to optimize the hyperparameters in the gradient boosting decision tree model using a hyperparameter optimization algorithm;

[0185] The training module 805 is used to train the optimized gradient boosting decision tree model using the training samples to obtain a dolomitization parameter data prediction model.

[0186] Corresponding to the above-mentioned dolomitization parameter data prediction method, some embodiments of this specification also provide a dolomitization parameter data prediction device, referring to Fig. 9 As shown, in some embodiments, the apparatus may include:

[0187] A receiving module 901 is used to receive a dolomitization parameter data set to be processed;

[0188] The prediction module 902 is used to input the dolomitization parameter data set to be processed into the dolomitization parameter data prediction model trained by the method described in any of the above embodiments, and predict the corresponding dolomitization parameter prediction data.

[0189] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0190] It should be noted that in the embodiments of this specification, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user and fully authorized by all parties.

[0191] It should be noted that the computer program product described in this specification is a software product that mainly implements the method described in this specification through a computer program.

[0192] The embodiments of this specification also provide a computer device. Fig.10As shown, in some embodiments of the present specification, the computer device 1002 may include one or more processors 1004, such as one or more central processing units (CPUs) or graphics processing units (GPUs), and each processing unit may implement one or more hardware threads. The computer device 1002 may also include any memory 1006, which is used to store any kind of information such as code, settings, data, etc. In a specific embodiment, the computer program on the memory 1006 and can be run on the processor 1004, when the computer program is run by the processor 1004, the instructions of the method described in any of the above embodiments may be executed. Non-limitingly, for example, the memory 1006 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 1002. In one embodiment, when the processor 1004 executes the associated instructions stored in any memory or combination of memories, the computer device 1002 can perform any operation of the associated instructions. The computer device 1002 also includes one or more drive mechanisms 1008 for interacting with any memory, such as a hard disk drive mechanism, an optical disk drive mechanism, etc.

[0193] The computer device 1002 may also include an input / output interface 1010 (I / O) for receiving various inputs (via input device 1012) and for providing various outputs (via output device 1014). A specific output mechanism may include a presentation device 1016 and an associated graphical user interface 1018 (GUI). In other embodiments, the input / output interface 1010 (I / O), input device 1012, and output device 1014 may not be included, and the computer device 1002 may be used as a computer device in a network. The computer device 1002 may also include one or more network interfaces 1020 for exchanging data with other devices via one or more communication links 1022. One or more communication buses 1024 couple the components described above together.

[0194] The communication link 1022 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 1022 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc. governed by any protocol or combination of protocols.

[0195] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), computer-readable storage media, and computer program products of some embodiments of the present specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processor to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processor generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0196] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processor to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0197] These computer program instructions can also be loaded onto a computer or other programmable data processor so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0198] In a typical configuration, a computer device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0199] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0200] Computer-readable media include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computer device. As defined in this specification, computer-readable media does not include temporary computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0201] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, the embodiments of this specification may take the form of complete hardware embodiments, complete software embodiments or embodiments combining software and hardware. Moreover, the embodiments of this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0202] The present specification embodiments may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present specification embodiments may also be practiced in distributed computing environments where tasks are performed by remote processors connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0203] It should also be understood that in the embodiments of this specification, the term "and / or" is only a description of the association relationship of the associated objects, indicating that three relationships can 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 this article generally indicates that the associated objects before and after are in an "or" relationship.

[0204] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0205] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiments of this specification.

[0206] In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples without contradiction.

[0207] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for establishing a dolomitization parameter data prediction model, characterized in that: The method comprises: Obtain dolomitization parameter data sets corresponding to dolomitization samples; Screening the dolomitization parameter data set to obtain a screening result; Constructing a training sample according to the screening results; Use hyperparameter optimization algorithms to optimize hyperparameters in the gradient boosting decision tree model; The training samples are used to train the optimized gradient boosting decision tree model to obtain a dolomitization parameter data prediction model.

2. The method according to claim 1, characterized in that Screening the dolomitization parameter data set to obtain the screening result further includes: According to the attribute characteristics of the dolomitization parameter data set, the dolomitization parameter data set is classified to obtain several classified data sets; different classified data sets include corresponding several attribute characteristic data sets.

3. The method according to claim 2, characterized in that The dolomitization parameter data set is screened to obtain screening results, including: Initialize the weights of all attribute feature data sets; From the dolomitization parameter data set, a property feature data set is selected by using a random algorithm as a central data set; From the classification data set where the central data set is located, a property feature data set is selected using a random algorithm as the first nearest neighbor data set; From the classification data sets other than the classification data set where the central data set is located, a property feature data set is selected by using a random algorithm as the second nearest neighbor data set; Calculate the first distance and the second distance between the central data set and the first nearest neighbor data set and the second nearest neighbor data set respectively; comparing the first distance and the second distance; If the first distance is not greater than the second distance, the weight of the central data set is updated using the weight gain coefficient; If the first distance is greater than the second distance, the weight of the central data set is updated using the weight decay coefficient; Repeat the steps of randomly selecting a central data set, a first nearest neighbor data set, and a second nearest neighbor data set, and updating the weight of the central data set according to the first distance and the second distance between the central data set and the first nearest neighbor data set and the second nearest neighbor data set, until a preset convergence condition is reached, thereby updating the weights of all attribute feature data sets; The attribute feature data set whose weight is less than a preset threshold is removed from the dolomitization parameter data set to obtain the screening result.

4. The method according to claim 1, characterized in that: The base learner of the gradient boosting decision tree model is a symmetric decision tree.

5. The method according to claim 1, characterized in that: Use hyperparameter optimization algorithms to optimize the hyperparameters in the gradient boosting decision tree model, including: Randomly initialize to obtain hyperparameter data to establish a corresponding optimization space; the hyperparameter data includes at least one or more of the number of iterations, learning rate, and maximum depth; Establish the objective function for hyperparameter optimization; In the optimization space, the objective function is subjected to Gaussian process processing to obtain a marginal likelihood distribution; Calculating the optimal solution of the current hyperparameters according to the marginal likelihood distribution; According to the current optimal solution of the hyperparameters, the objective function is optimized and updated using a preset acquisition function to obtain the optimal hyperparameters; The gradient boosting decision tree model is optimized according to the optimal hyperparameters.

6. The method according to claim 5, characterized in that Randomly initialize the hyperparameter data to establish the corresponding optimization space, including: In the preset mapping space, the distribution space of hyperparameters is randomly initialized using chaotic mapping; According to the preset number of data required, the hyperparameter data are selected from the distribution space, and each type of hyperparameter data conforms to the uniform distribution, so as to establish an optimization space according to the hyperparameter data.

7. The method according to claim 5, characterized in that The optimized gradient boosting decision tree model is trained using the training samples to obtain a dolomitization parameter data prediction model, including: Initializing a learner; the learner includes at least one decision tree; Using the training samples to train the learner to obtain a prediction result; According to the real labels and prediction results in the training samples, the loss is calculated using a preset loss function; Calculate the negative gradient based on the loss and the prediction result; Performing weighted random sampling on the training samples using the negative gradient to obtain sampled training samples; Using the sampled training samples to train a new decision tree; Scaling the trained new decision tree according to a preset learning rate, and adding the scaling result to the learner to update the learner; Repeat the steps of training a new decision tree using the sampled training samples and updating the learner according to the trained new decision tree until the number of iterations reaches a preset threshold.

8. A dolomitization parameter data prediction method, characterized in that: The method comprises: receiving a dolomitization parameter data set to be processed; The dolomitization parameter data set to be processed is input into a dolomitization parameter data prediction model trained by the method according to any one of claims 1 to 7 to predict corresponding dolomitization parameter prediction data.

9. A device for establishing a dolomitization parameter data prediction model, characterized in that: The device comprises: An acquisition module, used to acquire a dolomitization parameter data set corresponding to a dolomitization sample; A screening module, used for screening the dolomitization parameter data set to obtain a screening result; A construction module, used to construct a training sample according to the screening result; The optimization module is used to optimize the hyperparameters in the gradient boosting decision tree model using the hyperparameter optimization algorithm; The training module is used to train the optimized gradient boosting decision tree model using the training samples to obtain a dolomitization parameter data prediction model.

10. A dolomitization parameter data prediction device, characterized in that: The device comprises: A receiving module, used for receiving a dolomitization parameter data set to be processed; A prediction module is used to input the dolomitization parameter data set to be processed into a dolomitization parameter data prediction model trained by the method according to any one of claims 1 to 7, and predict corresponding dolomitization parameter prediction data.

Images