Liquid production profile prediction method, system and device based on physical constraint and storage medium

Abstract

The invention relates to a liquid production profile prediction method, system and device based on physical constraint and a storage medium, and the method comprises the following steps: determining influence factors of a horizontal well liquid production profile, collecting data, and obtaining influence parameter data; performing preprocessing and data division on the influence parameter data to obtain a training set and a test set; extracting and constructing prediction features; constructing a liquid production profile prediction model based on the prediction characteristics; training a fluid production profile prediction model according to the training set to obtain a trained fluid production profile prediction model; the trained liquid production profile model is optimized, and the trained liquid production profile prediction model is evaluated according to the test set; and predicting the liquid production profile of the horizontal well based on the evaluated liquid production profile prediction model to obtain a prediction result of the liquid production profile of the horizontal well. According to the method, horizontal well fluid production profile prediction based on physical constraint is achieved through the physical constraint fluid production profile prediction model, the prediction cost is low, the prediction precision is high, and the generalization ability is high.

Description

Technical Field

[0001] The present invention relates to the technical field of crude oil production, and more specifically, to a method, system, device and storage medium for predicting a liquid production profile based on physical constraints. Background Art

[0002] The conventional horizontal well production profile in offshore oilfield development and production is measured by mechanical water finding, numerical simulation software prediction method and output profile logging method. The mechanical water finding method is to test by sliding the casing downhole. It has low operating cost and simple process, but it can only be applied to wells with segmented wellbore and has a small scope of application. The output profile logging method requires the production of logging instruments under the drilling platform for operation. It can intuitively display the production profile and has high accuracy, but the testing cost is high, and most horizontal wells are difficult to meet the operating conditions.

[0003] The prediction method of numerical simulation software is based on geological models and uses the method of reservoir numerical simulation for prediction. This method can study heterogeneous and multiphase flow problems, but it is greatly affected by the grid scale and the accuracy of model history fitting. In recent years, there has been a significant increase in the research on deep learning technology for oil and gas production prediction. Among these methods, the most widely used algorithm is the time series-RNN (recurrent neural network) algorithm. However, traditional data-driven neural networks are limited by the size of data samples, have limited generalization ability, lack physical interpretability, and are difficult to meet the accuracy requirements of liquid production profile prediction. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method, system, device and storage medium for predicting liquid production profile based on physical constraints in view of the problems existing in the prior art.

[0005] The technical solution adopted by the present invention to solve the technical problem is: constructing a liquid production profile prediction method based on physical constraints, comprising the following steps:

[0006] Determine the factors affecting the fluid production profile of horizontal wells and collect data to obtain influencing parameter data;

[0007] Preprocessing and dividing the influencing parameter data to obtain a training set and a test set;

[0008] Extract and construct predictive features;

[0009] Building a neural network model based on the prediction features;

[0010] Training the neural network model according to the training set to obtain a trained liquid production profile prediction model;

[0011] Optimizing the trained fluid production profile model, and evaluating the trained fluid production profile prediction model according to the test set;

[0012] The fluid production profile of the horizontal well is predicted based on the fluid production profile prediction model after the evaluation, and the prediction result of the fluid production profile of the horizontal well is obtained.

[0013] In the method for predicting the production profile of a liquid well based on physical constraints described in the present invention, determining the influencing factors of the production profile of a horizontal well and collecting data to obtain the influencing parameter data includes:

[0014] The grey correlation method and the range analysis method are used to analyze and determine the influencing factors of the horizontal well fluid production profile;

[0015] Data is collected based on the influencing factors of the horizontal well fluid production profile to obtain the influencing parameter data.

[0016] In the method for predicting the production profile based on physical constraints of the present invention, the data collection based on the influencing factors of the production profile of the horizontal well to obtain the influencing parameter data includes:

[0017] Based on the influencing factors of the horizontal well production profile, Yaopai Latin hypercube sampling randomly generates different types of reservoir numerical models;

[0018] Based on the reservoir numerical model, reservoir numerical simulation software is used to simulate the production of horizontal wells to obtain the liquid production, water content data, total water content and total liquid production on the horizontal section profile of the horizontal well; the liquid production, water content data, total water content and total liquid production on the horizontal section profile of the horizontal well are the influencing parameter data.

[0019] In the method for predicting liquid production profile based on physical constraints described in the present invention, constructing a neural network model based on the prediction features includes:

[0020] A neural network with an encoding-decoding structure is used to construct the neural network model, and the input of the neural network model is the prediction feature, and the output of the neural network model is the liquid production profile of the horizontal section.

[0021] In the method for predicting liquid production profile based on physical constraints described in the present invention, the neural network model includes: a sequence feature encoder, a single well parameter feature encoder and a decoder;

[0022] The sequence feature encoder includes a convolutional neural network or a multi-layer perceptron;

[0023] The single well parameter feature encoder includes a multi-layer perceptron and a gated linear unit;

[0024] The decoder includes a multi-layer perceptron.

[0025] In the method for predicting a liquid production profile based on physical constraints of the present invention, the training of the neural network model according to the training set to obtain a trained liquid production profile prediction model includes:

[0026] Based on the training set, the neural network model is trained using a gradient descent algorithm to obtain the trained liquid production profile prediction model.

[0027] In the method for predicting a liquid production profile based on physical constraints of the present invention, the step of optimizing the trained liquid production profile model and evaluating the trained liquid production profile prediction model according to the test set includes:

[0028] Optimizing the trained fluid production profile model by adjusting the hyperparameters of the trained fluid production profile model;

[0029] After the optimization of the trained fluid production profile model is completed, the trained fluid production profile prediction model is evaluated based on the test set.

[0030] The present invention also provides a liquid production profile prediction system based on physical constraints, comprising:

[0031] A data collection unit, used to determine the influencing factors of the horizontal well fluid production profile and collect data to obtain influencing parameter data;

[0032] A data preprocessing unit, used to preprocess and divide the influencing parameter data to obtain a training set and a test set;

[0033] A feature extraction unit, used to extract and construct prediction features;

[0034] A model building unit, used for building a neural network model based on the prediction features;

[0035] A model training unit, used for training the liquid production profile prediction model according to the training set to obtain a trained liquid production profile prediction model;

[0036] A model optimization and evaluation unit, used for optimizing the trained fluid production profile model and evaluating the trained fluid production profile prediction model according to the test set;

[0037] The prediction unit is used to predict the fluid production profile of the horizontal well based on the fluid production profile prediction model after the evaluation, and obtain the prediction result of the fluid production profile of the horizontal well.

[0038] The present invention also provides a storage medium storing a computer program, wherein the computer program is suitable for being loaded by a processor to execute the steps of the liquid production profile prediction method based on physical constraints as described above.

[0039] The present invention also provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the steps of the above-mentioned method for predicting liquid production profile based on physical constraints by calling the computer program stored in the memory.

[0040] The implementation of the method, system, device and storage medium for predicting the production profile based on physical constraints of the present invention has the following beneficial effects: including the following steps: determining the influencing factors of the production profile of the horizontal well and collecting data to obtain the influencing parameter data; preprocessing the influencing parameter data and data partitioning to obtain the training set and the test set; extracting and constructing the prediction features; building a neural network model based on the prediction features; training the neural network model according to the training set to obtain the trained production profile prediction model; optimizing the trained production profile model and evaluating the trained production profile prediction model according to the test set; predicting the production profile of the horizontal well based on the production profile prediction model after the evaluation to obtain the prediction result of the production profile of the horizontal well. The present invention realizes the prediction of the production profile of the horizontal well based on the physical constraints through the production profile prediction model of the physical constraints, with low prediction cost, high prediction accuracy and strong generalization ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0042] Figure 1 It is a schematic flow chart of the method for predicting liquid production profile based on physical constraints provided by the present invention;

[0043] Figure 2 It is a schematic diagram of the framework of the physical constraint liquid production profile prediction model provided by the present invention;

[0044] Figure 3 It is a functional block diagram of a liquid production profile prediction system based on physical constraints provided by the present invention;

[0045] Figure 4 It is a schematic diagram of the prediction result of the liquid production of Well 1 provided by the present invention;

[0046] Figure 5 It is a schematic diagram of the prediction result of the liquid production of Well 2 provided by the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] In order to solve the problem of existing horizontal well production profile prediction, the present invention provides a horizontal well production profile prediction model based on physical constraints. First, determine the influencing factors of the horizontal well production profile; establish a mechanism model through reservoir numerical simulation software (such as Eclipse) to generate horizontal well production profile data; determine the single well parameters and sequence parameters of the model production profile output, and divide the training set and the test set; establish a data-driven production profile prediction model with convolutional neural network and fully connected neural network as basic units; establish a physical constraint loss function composed of the reservoir-horizontal well coupling equation loss term and the data mean square error (MSE) loss term; output the production profile data, determine the comprehensive error size through the physical constraint loss function, train the physical constraint production profile prediction neural network, optimize the physical constraint loss function through the gradient descent algorithm, find the minimum value of the comprehensive error to obtain the optimal grid weight, and then realize the horizontal well production profile prediction based on physical constraints.

[0049] Specifically, Figure 1 As shown, the liquid production profile prediction method based on physical constraints includes the following steps:

[0050] Step S101: determine the influencing factors of the horizontal well fluid production profile and collect data to obtain influencing parameter data.

[0051] In some embodiments, determining the influencing factors of the horizontal well fluid production profile and collecting data to obtain the influencing parameter data includes: using the grey correlation method and the range analysis method to perform analysis to determine the influencing factors of the horizontal well fluid production profile; and collecting data based on the influencing factors of the horizontal well fluid production profile to obtain the influencing parameter data.

[0052] Among them, data collection is performed based on the influencing factors of the horizontal well production profile, and the influencing parameter data are obtained, including: based on the influencing factors of the horizontal well production profile, waist brand Latin hypercube sampling randomly generates different types of reservoir numerical models; based on the reservoir numerical model, the reservoir numerical simulation software is used to simulate the horizontal well production, and the liquid production, water content data, total water content and total liquid production of the horizontal well on the horizontal section profile of the horizontal well are obtained. It should be noted that the liquid production, water content data, total water content and total liquid production of the horizontal well on the horizontal section profile of the horizontal well are influencing parameter data. In the embodiment of the present invention, the influencing factors include but are not limited to: interlayer distribution, distance from the interlayer to the horizontal section, permeability difference between sections, permeability along the way, water avoidance height / edge water distance, oil phase viscosity, total liquid production, horizontal section length, total water content, water body energy, etc.

[0053] Step S102: pre-processing and data division of the influencing parameter data to obtain a training set and a test set.

[0054] Specifically, after obtaining the influencing parameter data, all data are divided into a training set and a test set in a ratio of 7:3. Among them, the maximum and minimum values ​​of each input parameter in the training set are used as the standard, and the Min-Max normalization is used to normalize the data set.

[0055] Step S103: extracting and constructing prediction features.

[0056] In some embodiments, features suitable for prediction, i.e., prediction features, can be extracted and constructed based on knowledge in the field of seepage mechanics and data characteristics. Among them, prediction features include: sequence features and single well parameter features. Sequence features are features that change with spatial position, such as the distance of the horizontal well section from the oil-water interface, the permeability along the horizontal well, and the position of the interlayer along the layer and the height from the horizontal well. Single well parameter features are single-point features that remain unchanged at a certain moment in the production of the oil well, such as oil phase viscosity, bottom water energy, total water content, and other features.

[0057] Step S104: constructing a neural network model based on the prediction features.

[0058] In some embodiments, constructing a neural network model based on prediction features includes: using a neural network with an encoding-decoding structure to construct a neural network model, and the input of the neural network model is the prediction feature, and the output of the neural network model is the liquid production profile of the horizontal section.

[0059] Among them, the neural network model includes: a sequence feature encoder, a single well parameter feature encoder and a decoder; the sequence feature encoder includes a convolutional neural network or a multi-layer perceptron; the single well parameter feature encoder includes a multi-layer perceptron and a gated linear unit; the decoder includes a multi-layer perceptron.

[0060] Specifically, Figure 2 As shown, the liquid production profile prediction model described in the embodiment of the present invention is a physically constrained liquid production profile prediction model, which is a neural network with an encoding-decoding structure. Among them, the encoder is divided into a sequence feature encoder and a single well parameter feature encoder. The sequence feature encoder adopts a convolutional neural network (CNN) or a multi-layer perceptron (MLP) as the basic framework, and the single well parameter feature encoder adopts a multi-layer perceptron (MLP) and a gated linear unit (GLU) as the basic framework. The decoder adopts a multi-layer perceptron (MLP) as the basic framework to construct a physical constraint loss function composed of the residual loss term and the data matching loss term of the reservoir-horizontal well coupling equation, wherein the input of the neural network model is the single well parameter characteristics of a horizontal well and the sequence characteristics of the horizontal section, and the output is the liquid production profile data of the horizontal section.

[0061] Step S105: training the neural network model according to the training set to obtain a trained liquid production profile prediction model.

[0062] In some embodiments, training the neural network model according to the training set to obtain the trained liquid production profile prediction model includes: based on the training set, using a gradient descent algorithm to train the neural network model to obtain the trained liquid production profile prediction model.

[0063] Specifically, according to the data in the training set in step S102 and the neural network model constructed in step S104, the physical constraint loss function composed of the residual loss term and the data matching loss term of the reservoir-horizontal well coupling equation is optimized by using the gradient descent algorithm to obtain the minimum value of the loss function and obtain the trained liquid production profile prediction model. Among them, the physical constraint loss function is used to optimize the neural network model (i.e., the objective function of the neural network model constructed in step S104). The trained liquid production profile prediction model is the neural network model obtained after training.

[0064] Step S106: Optimize the trained liquid production profile model, and evaluate the trained liquid production profile prediction model based on the test set.

[0065] In some embodiments, the trained fluid production profile model is optimized, and the trained fluid production profile prediction model is evaluated based on a test set, including: optimizing the trained fluid production profile model by adjusting the hyperparameters of the trained fluid production profile model; after completing the optimization of the trained fluid production profile model, evaluating the trained fluid production profile prediction model based on the test set.

[0066] Specifically, the hyperparameters of the neural network model of the encoding-decoding structure, such as the network structure, learning rate, batch size, etc., can be adjusted as needed to optimize the performance and prediction accuracy of the model and optimize the trained liquid production profile prediction model. Then, the trained liquid production profile prediction model is evaluated using the data of the divided test set to evaluate the prediction performance of the model. Among them, various evaluation indicators, such as root mean square error (RMSE), mean absolute error (MAE), etc., can be used to evaluate the prediction performance of the model.

[0067] Step S107: predicting the fluid production profile of the horizontal well based on the evaluated fluid production profile prediction model to obtain a prediction result of the fluid production profile of the horizontal well.

[0068] Specifically, after completing the evaluation of the fluid production profile prediction model, the trained model is used to predict the fluid production profile of the horizontal wells in the test set. The fluid production profiles of horizontal wells with different sequence characteristics and different single well parameter characteristics can be predicted as needed. The model prediction results are analyzed to evaluate the accuracy and stability of the model.

[0069] refer to Figure 3 , Figure 3 This is a functional block diagram of the liquid production profile prediction system based on physical constraints provided by the present invention.

[0070] Specifically, Figure 3 As shown, the liquid production profile prediction system based on physical constraints includes:

[0071] The data collection unit 301 is used to determine the influencing factors of the horizontal well fluid production profile and collect data to obtain influencing parameter data.

[0072] The data preprocessing unit 302 is used to preprocess the influencing parameter data and divide the data to obtain a training set and a test set.

[0073] The feature extraction unit 303 is used to extract and construct prediction features.

[0074] The model building unit 304 is used to build a neural network model based on the prediction features.

[0075] The model training unit 305 is used to train the liquid production profile prediction model according to the training set to obtain a trained liquid production profile prediction model.

[0076] The model optimization and evaluation unit 306 is used to optimize the trained liquid production profile model and evaluate the trained liquid production profile prediction model according to the test set.

[0077] The prediction unit 307 is used to predict the liquid production profile of the horizontal well based on the evaluated liquid production profile prediction model to obtain a prediction result of the liquid production profile of the horizontal well.

[0078] Specifically, the specific coordination operation process between the various units in the liquid production profile prediction system based on physical constraints can refer to the above-mentioned liquid production profile prediction method based on physical constraints, which will not be repeated here.

[0079] In order to verify the effect of using the physical constraint liquid production profile prediction model to predict the liquid production profile of horizontal wells in bottom water reservoirs, the present invention uses ECLIPSE reservoir numerical simulation software to simulate the mechanism model of 5000 bottom water oil horizontal wells to obtain the corresponding input characteristic data and the liquid production along the way, production pressure difference and water saturation data. The details are as follows:

[0080] Firstly, the liquid production, production pressure difference and water saturation data of the mechanism model as well as the relevant single well parameter characteristics and along-process sequence characteristic data are collected and sorted.

[0081] Next, the data set is divided into training set and test set in a ratio of 7:3. The data is preprocessed.

[0082] However, based on the knowledge of seepage mechanics and data characteristics, features suitable for prediction are extracted and constructed. These features can include the distance of the horizontal well section from the oil-water interface, the permeability along the horizontal well, and the position of the interlayer along the way and the height from the horizontal well.

[0083] Next, a physically constrained liquid production profile prediction model is built. The model is a neural network with an encoding-decoding structure. The encoder is divided into a sequence feature encoder and a single well parameter feature encoder. The sequence feature encoder uses CNN or MLP as the basic framework, and the single well parameter feature encoder is constructed using MLP and GLU as the basic framework. The decoder uses MLP as the basic framework to construct a physical constraint loss function composed of the residual loss term and the data matching loss term of the reservoir-horizontal well coupling equation. The network input is the single well parameters of a horizontal well and the sequence characteristics of the horizontal section, and the output is the liquid production profile of the horizontal section.

[0084] The physical constraint liquid production profile prediction model established adopts a convolutional neural network algorithm and a loss function that considers physical constraints. The algorithm is a sequence prediction model based on a convolutional neural network (CNN). The input of the model includes single well parameter data and sequence feature data corresponding to the horizontal well sequence. The model predicts the liquid production profile under other single well parameters and sequence features by learning the single well parameters and sequence features of the horizontal wells in the training set.

[0085] Some parameter settings of the model are as follows:

[0086] Single well parameter characteristics: represents the non-sequential parameters of a horizontal well in a bottom water reservoir, such as the length of the horizontal well, oil phase viscosity, bottom water energy, total liquid production of the horizontal well, and total water content of the horizontal well.

[0087] Sequence parameter characteristics: represents the sequence parameters of a horizontal well in a bottom water reservoir, such as the permeability along the horizontal well, the height information of the horizontal well section, the lateral distribution and vertical distribution position of the interlayer, etc.

[0088] Single well parameter feature encoder: used to extract single well parameter features. MLP and GLU are used as basic units. Here, the basic forward neural network framework is defined, including the number of network layers and the number of basic units in each layer (both of which are hyperparameters that need to be optimized).

[0089] Sequence parameter feature encoder: used to extract sequence parameter features. It is composed of MLP and CONV (convolutional neural network) as basic units. Here we define the basic forward neural network framework, including the number of network layers and the number of basic units in each layer (both of which are hyperparameters that need to be optimized).

[0090] Decoder: used to decode the code and predict the liquid production profile. MLP is the basic unit. Here we define the basic forward neural network framework, including the number of network layers and the number of basic units in each layer (both of which are hyperparameters that need to be optimized).

[0091] MSE phy Represents the physical constraint term. The residual of the reservoir-horizontal well coupling equation is in the form of mean square error.

[0092] Among them, MSE phy The specific expression is as follows:

[0093]

[0094] Among them, x i Represents the sequence features in the neural network input parameters; k i represents the permeability of the i-th horizontal segment; L i represents the length of the i-th horizontal segment; r e Represents the equivalent radius of the horizontal well; r w Indicates the radius of the horizontal wellbore; Δp i represents the production pressure difference of the i-th horizontal section; k rw and k ro Represent the relative permeabilities of water phase and oil phase respectively; μ w and μ o Represent the viscosity of water phase and oil phase respectively; B w and Bo represent the volume coefficients of water phase and oil phase respectively.

[0095] MSE data Represents the data constraint term. It is the mean square error form between the true value and the neural network prediction value. Among them, MSE phy The specific expression is as follows:

[0096]

[0097] Among them, q li is the liquid production rate of the i-th segment of the horizontal section, m 3 / d; Net(·) represents the output of the neural network; x i represents the sequence features in the neural network input parameters; s represents the single well parameters in the neural network input parameters.

[0098] The comprehensive physical constraint loss function Loss can be expressed as:

[0099] Loss = λ phy MSE phy +λ data MSE data .

[0100] Among them, λ represents the weight of the constraint item, which is generally 0.5.

[0101] The gradient descent algorithm is used to optimize the loss function of the physical constraints, find the minimum value of the loss function, and obtain the liquid production profile. As needed, the hyperparameters of the neural network model, such as network structure, learning rate, batch size, etc., can be adjusted to optimize the performance and prediction accuracy of the model. The trained model is evaluated using the test set. Various evaluation indicators, such as root mean square error (RMSE), mean absolute error (MAE), etc., can be used to evaluate the prediction performance of the model. Figure 4 and Figure 5 It is the prediction result and relative error of the production profile of two horizontal wells under different conditions by the physically constrained production prediction model.

[0102] As can be seen from the figure, the physical constraint liquid production profile prediction model can accurately predict the liquid production profile, the output results are physically interpretable, conform to the physical laws of seepage, and have strong generalization ability. The average determination coefficient R2 on the prediction set of more than 3,000 wells is 0.8, indicating that the prediction performance of the liquid production profile prediction model considering physical constraints is very good.

[0103] The present invention uses the reservoir-horizontal well coupling equation as a regularized physical constraint term of the loss function, and uses the constraint neural network of the multi-objective loss function to train the production profile prediction model. A codec structure neural network with a convolutional neural network CNN+fully connected neural network MLP as an encoder and a multi-layer MLP as a decoder is constructed, which can simultaneously extract single well features and sequence features to achieve variable sequence length prediction of the production profile.

[0104] In addition, an electronic device of the present invention includes a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program to implement a liquid production profile prediction method based on physical constraints as described in any one of the above. Specifically, according to an embodiment of the present invention, the process described with reference to the flowchart above can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed by an electronic device and when executed, it performs the above functions defined in the method of the embodiment of the present invention. The electronic device in the present invention can be a terminal such as a notebook, a desktop, a tablet computer, a smart phone, or a server.

[0105] In addition, a storage medium of the present invention stores a computer program, and when the computer program is executed by a processor, it implements any of the above methods for predicting liquid production profiles based on physical constraints. Specifically, it should be noted that the above storage medium of the present invention can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it. In the present invention, a computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer readable signal media may also be any computer readable medium other than computer readable storage media, which may send, propagate, or transmit programs for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0106] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0107] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0108] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in 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 the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may 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.

[0109] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0110] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot limit the scope of protection of the present invention. All equivalent changes and modifications made to the scope of the claims of the present invention should fall within the scope of the claims of the present invention.

Claims

1. A method for predicting liquid production profile based on physical constraints, characterized in that: The following steps are involved: Determine the factors affecting the fluid production profile of horizontal wells and collect data to obtain influencing parameter data; Preprocessing and dividing the influencing parameter data to obtain a training set and a test set; Extract and construct predictive features; Building a neural network model based on the prediction features; Training the neural network model according to the training set to obtain a trained liquid production profile prediction model; Optimizing the trained fluid production profile model, and evaluating the trained fluid production profile prediction model according to the test set; The fluid production profile of the horizontal well is predicted based on the fluid production profile prediction model after the evaluation, and the prediction result of the fluid production profile of the horizontal well is obtained.

2. The method for predicting liquid production profile based on physical constraints according to claim 1, characterized in that: Determining the influencing factors of the horizontal well fluid production profile and collecting data to obtain influencing parameter data includes: The grey correlation method and the range analysis method are used to analyze and determine the influencing factors of the horizontal well fluid production profile; Data is collected based on the influencing factors of the horizontal well fluid production profile to obtain the influencing parameter data.

3. The method for predicting liquid production profile based on physical constraints according to claim 2, characterized in that: The data collection based on the influencing factors of the horizontal well fluid production profile to obtain the influencing parameter data includes: Based on the influencing factors of the horizontal well production profile, Yaopai Latin hypercube sampling randomly generates different types of reservoir numerical models; Based on the reservoir numerical model, reservoir numerical simulation software is used to simulate the production of horizontal wells to obtain the liquid production, water content data, total water content and total liquid production on the horizontal section profile of the horizontal well; the liquid production, water content data, total water content and total liquid production on the horizontal section profile of the horizontal well are the influencing parameter data.

4. The method for predicting liquid production profile based on physical constraints according to claim 1, characterized in that: The constructing of a neural network model based on the prediction features comprises: A neural network with an encoding-decoding structure is used to construct the neural network model, and the input of the neural network model is the prediction feature, and the output of the neural network model is the liquid production profile of the horizontal section.

5. The method for predicting liquid production profile based on physical constraints according to claim 4, characterized in that: The neural network model includes: a sequence feature encoder, a single well parameter feature encoder and a decoder; The sequence feature encoder includes a convolutional neural network or a multi-layer perceptron; The single well parameter feature encoder includes a multi-layer perceptron and a gated linear unit; The decoder includes a multi-layer perceptron.

6. The method for predicting liquid production profile based on physical constraints according to claim 1, characterized in that: The step of training the neural network model according to the training set to obtain a trained liquid production profile prediction model comprises: Based on the training set, the neural network model is trained using a gradient descent algorithm to obtain the trained liquid production profile prediction model.

7. The method for predicting liquid production profile based on physical constraints according to claim 1, characterized in that: The optimizing the trained fluid production profile model and evaluating the trained fluid production profile prediction model according to the test set includes: Optimizing the trained fluid production profile model by adjusting the hyperparameters of the trained fluid production profile model; After the optimization of the trained fluid production profile model is completed, the trained fluid production profile prediction model is evaluated based on the test set.

8. A liquid production profile prediction system based on physical constraints, characterized in that: include: A data collection unit, used to determine the influencing factors of the horizontal well fluid production profile and collect data to obtain influencing parameter data; A data preprocessing unit, used to preprocess and divide the influencing parameter data to obtain a training set and a test set; A feature extraction unit, used to extract and construct prediction features; A model building unit, used for building a neural network model based on the prediction features; A model training unit, used to train the neural network model according to the training set to obtain a trained liquid production profile prediction model; A model optimization and evaluation unit, used for optimizing the trained fluid production profile model and evaluating the trained fluid production profile prediction model according to the test set; The prediction unit is used to predict the fluid production profile of the horizontal well based on the fluid production profile prediction model after the evaluation, and obtain the prediction result of the fluid production profile of the horizontal well.

9. A storage medium, characterized in that: The storage medium stores a computer program, and the computer program is suitable for being loaded by a processor to execute the steps of the method for predicting a liquid production profile based on physical constraints as described in any one of claims 1 to 7.

10. An electronic device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program, and the processor executes the steps of the method for predicting a liquid production profile based on physical constraints as described in any one of claims 1 to 7 by calling the computer program stored in the memory.

Images