Production operation state optimization method of oil field gathering and transportation system, related method, device and equipment
By applying a one-dimensional convolutional neural network prediction model in the oil field collection and transmission system, the problem of insufficient real-time and accuracy of traditional manual monitoring is solved, real-time optimization and energy saving of the oil field collection and transmission system are achieved, and production efficiency and reliability are improved. sex.
Patent Information
- Application Number
- CN202310506224.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-06
- Publication Date
- 2024-11-08
AI Technical Summary
Traditional oilfield collection and transportation systems rely on manual monitoring and empirical judgment, resulting in poor real-time, insufficient accuracy and low reliability, making it difficult to quickly respond to production changes and optimize the oilfield production environment.
A prediction model based on one-dimensional convolutional neural network is adopted to obtain historical operation data for preprocessing and training, predict parameters such as pump station pressure, temperature and flow, calculate parameter errors and dynamically adjust the system status to achieve real-time optimization.
It improves the real-time monitoring and optimization capabilities of the oilfield collection and transportation system, reduces human errors and energy consumption, enhances production efficiency and reliability, and adapts to the complex and changeable oilfield production environment.
Smart Images

Figure CN118917162A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a production operation state optimization method of an oilfield gathering and transportation system, and related methods, devices and equipment. Background Art
[0002] In the process of oilfield exploitation, the gathering and transportation system is the key link between oil and gas from the wellhead to the production and processing facilities. Therefore, effectively controlling and optimizing the production and operation status of the gathering and transportation system is of great significance for improving oilfield production efficiency, reducing energy consumption and protecting the environment. Traditional oilfield gathering and transportation systems mainly rely on manual monitoring and experience judgment for operation adjustments. In recent years, the oilfield production field has begun to try to introduce artificial intelligence technology, using methods such as data mining and pattern recognition to realize intelligent production control and optimization of oilfield technical systems. At present, the research and application of artificial intelligence for oilfield gathering and transportation systems is still in its early stages. Summary of the invention
[0003] In order to better optimize the production and operation status of an oilfield gathering and transportation system, an embodiment of the present invention provides a production and operation status optimization method, related methods, devices and equipment of an oilfield gathering and transportation system.
[0004] In a first aspect, an embodiment of the present invention provides a method for optimizing the production and operation status of an oilfield gathering and transportation system, the method comprising:
[0005] Obtain at least one operating parameter within a preset historical time period, and use a pre-trained one-dimensional convolutional neural network-based prediction model to perform predictions to obtain at least one corresponding predicted operating parameter result;
[0006] Determining an operating parameter error between each of the predicted operating parameter results and a corresponding preset target operating parameter result;
[0007] At least one optimized operating parameter of the oilfield gathering and transportation system is determined according to each obtained operating parameter error and the corresponding current operating parameter, so as to dynamically adjust the real-time operating status of the oilfield gathering and transportation system.
[0008] Based on the foregoing implementation, in an optional implementation of the embodiment of the present application, the predicted operating parameter results include predicted pump station pressure results, predicted temperature results and predicted flow results; the preset target operating parameter results include preset target pump station pressure, preset target temperature and preset target flow; the operating parameter errors include pump station pressure error, temperature error and flow error;
[0009] The determining of the operating parameter error between each of the predicted operating parameter results and the corresponding preset target operating parameter result includes:
[0010] Based on the following formula 1, the pump station pressure error is calculated:
[0011] ΔP = P _target -P _pred Formula 1;
[0012] Where ΔP is the pressure error of the pump station; P _target is the preset target pump station pressure; P _pred The predicted pump station pressure result;
[0013] Based on the following formula 2, the temperature error is calculated:
[0014] ΔT = T _target -T _pred Formula 2;
[0015] Where ΔT is the temperature error; T _target is the preset target temperature; T _pred is the predicted temperature result;
[0016] Based on the following formula 3, the flow error is calculated:
[0017] ΔQ = Q _target -Q _pred Formula 3;
[0018] Where ΔQ is the flow error; Q _target is the preset target flow rate; Q _pred This is the predicted traffic result.
[0019] Based on the foregoing implementation, in an optional implementation of the embodiment of the present application, the current operating parameters include the current pump station pressure, the current temperature and the current flow; the optimized operating parameter results include the optimized pump station pressure, the optimized temperature and the optimized flow;
[0020] Determining at least one optimized operating parameter of the oilfield gathering and transportation system according to each obtained operating parameter error and the corresponding current operating parameter to dynamically adjust the real-time operating state of the oilfield gathering and transportation system includes:
[0021] Based on the following formula 4, the optimized pump station pressure is calculated:
[0022] P _new = P _current + k1 * ΔP Formula 4;
[0023] Where P _new is the pump station pressure after optimization; P _current is the current pump station pressure; k1 is the first preset adjustment coefficient; ΔP is the pump station pressure error;
[0024] Based on the following formula 5, the optimized temperature is calculated:
[0025] T _new = T _current + k2 * ΔT Formula 5;
[0026] Where, T _new is the optimized temperature; T _current is the current temperature; k2 is the second preset adjustment coefficient; ΔT is the temperature error;
[0027] Based on the following formula 6, the optimized flow is calculated:
[0028] Q _new = Q _current + k3 * ΔQ Formula 6;
[0029] In the formula, Q _new is the optimized flow rate; Q _current is the current flow rate; k3 is the third preset adjustment coefficient; ΔQ is the flow rate error;
[0030] The real-time operation status of the oilfield gathering and transportation system is dynamically adjusted according to the optimized pump station pressure, optimized temperature and optimized flow.
[0031] Based on the foregoing implementation, in an optional implementation of the embodiment of the present application, the prediction model is obtained by training in the following manner:
[0032] Dividing the previously acquired historical operation data of the oilfield gathering and transportation system into multiple training sets and multiple test sets;
[0033] Input each of the training sets into a pre-built one-dimensional convolutional neural network model for training;
[0034] Input each of the test sets into the trained one-dimensional convolutional neural network model for testing to obtain a test loss value;
[0035] Repeat the above model training and testing steps to obtain the one-dimensional convolutional neural network model with the smallest test loss value as the prediction model.
[0036] Based on the foregoing implementation, in an optional implementation of the embodiment of the present application, the step of inputting each of the training sets into a pre-built one-dimensional convolutional neural network model for training includes:
[0037] Input each of the training sets into a pre-built one-dimensional convolutional neural network model to obtain corresponding prediction training results;
[0038] Determine the loss value between each of the predicted training results and the corresponding true label result;
[0039] Determine the gradient of the model parameters corresponding to the loss value, and update the model parameters according to the gradient of the model parameters to obtain a trained one-dimensional convolutional neural network model.
[0040] Based on the foregoing implementation, in an optional implementation of the embodiment of the present application, the gradient of the model parameter includes a first-order moment estimation and a second-order moment estimation; according to the gradient of the model parameter, the model parameter is updated in the following manner:
[0041] Based on the following formula 7, use the Adam optimizer to update the parameters:
[0042] θ _t+1 = θ _t - learning_rate * (m _t / (sqrt(v _t ) + ε)) Formula 7;
[0043] In the formula, θ _t is the model parameter; m _t is the first-order moment estimate; v _t is the second-order moment estimate; learning_rate is the learning rate; ε is a small constant to prevent division by zero.
[0044] Based on the foregoing implementation, in an optional implementation of the embodiment of the present application, before dividing the pre-acquired historical operation data of the oilfield gathering and transportation system into a training set and a test set, it also includes:
[0045] The historical operation data of the oilfield gathering and transportation system acquired in advance are preprocessed; the preprocessing method includes at least one of normalization, noise reduction and feature selection.
[0046] Based on the foregoing implementation, an optional implementation of the embodiment of the present application further includes:
[0047] Acquire at least one real-time operating parameter collected during the operation of the oilfield gathering and transportation system within a preset operating time period;
[0048] Retraining the prediction model according to at least one real-time operating parameter within the preset operating time period, and updating the model parameters using a stochastic gradient descent algorithm to obtain an updated prediction model;
[0049] The updated prediction model is used to perform prediction to obtain at least one corresponding prediction operation parameter result.
[0050] In a second aspect, an embodiment of the present invention provides a training method for a prediction model based on a one-dimensional convolutional neural network, the method comprising:
[0051] Dividing the previously acquired historical operation data of the oilfield gathering and transportation system into multiple training sets and multiple test sets;
[0052] Input each of the training sets into a pre-built one-dimensional convolutional neural network model for training;
[0053] Input each of the test sets into the trained one-dimensional convolutional neural network model for testing to obtain a test loss value;
[0054] Repeat the above model training and testing steps to obtain the one-dimensional convolutional neural network model with the smallest test loss value as the prediction model.
[0055] In a third aspect, an embodiment of the present invention provides a production operation state optimization device for an oilfield gathering and transportation system, the device comprising:
[0056] A prediction parameter module is used to obtain at least one operating parameter within a preset historical time period, and use a pre-trained prediction model based on a one-dimensional convolutional neural network to make a prediction to obtain at least one corresponding predicted operating parameter result;
[0057] An error determination module is used to determine an operating parameter error between each of the predicted operating parameter results and a corresponding preset target operating parameter result;
[0058] The parameter optimization module is used to determine at least one optimized operating parameter of the oilfield gathering and transportation system according to each obtained operating parameter error and the corresponding current operating parameter, so as to dynamically adjust the real-time operating state of the oilfield gathering and transportation system.
[0059] In a fourth aspect, an embodiment of the present invention provides a training device for a prediction model based on a one-dimensional convolutional neural network, the device comprising:
[0060] The first training module is used to divide the pre-acquired historical operation data of the oilfield gathering and transportation system into multiple training sets and multiple test sets;
[0061] A second training module is used to input each of the training sets into a pre-built one-dimensional convolutional neural network model for training;
[0062] A third training module is used to input each of the test sets into the trained one-dimensional convolutional neural network model for testing to obtain a test loss value;
[0063] The fourth training module is used to repeat the above model training and testing steps to obtain the one-dimensional convolutional neural network model with the smallest test loss value as the prediction model.
[0064] In a fifth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the production operation status optimization method of the oil field gathering and transportation system as described above and / or the training method of a prediction model based on a one-dimensional convolutional neural network.
[0065] In a sixth aspect, an embodiment of the present invention provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the production operation status optimization method of the oilfield gathering and transportation system as described above and / or the training method of the prediction model based on the one-dimensional convolutional neural network.
[0066] In the seventh aspect, an embodiment of the present invention provides a computer program product comprising instructions. When the computer program product is run on a computer device, the computer device executes the production operation status optimization method of the oil field gathering and transportation system and / or the training method of the prediction model based on the one-dimensional convolutional neural network as mentioned above.
[0067] In an eighth aspect, an embodiment of the present invention provides a chip, comprising a processor and a communication interface, wherein the communication interface and the processor are coupled, and the processor is used to run a computer program or instructions to implement the production operation status optimization method of the oil field gathering and transportation system as described above and / or the training method of a prediction model based on a one-dimensional convolutional neural network.
[0068] The beneficial effects of the above technical solution provided by the embodiment of the present invention include at least:
[0069] The production operation state optimization method of the oilfield gathering and transportation system provided by the embodiment of the present invention obtains at least one operation parameter in a preset historical time period, and uses a pre-trained prediction model based on a one-dimensional convolutional neural network to predict, obtains at least one corresponding predicted operation parameter result, then determines the operation parameter error between each predicted operation parameter result and the corresponding preset target operation parameter result, and finally, according to each obtained operation parameter error and the corresponding current operation parameter, determines at least one optimized operation parameter of the oilfield gathering and transportation system to dynamically adjust the real-time operation state of the oilfield gathering and transportation system. Compared with the prior art, the method can quickly respond to changes in the oilfield production environment through real-time monitoring of the prediction model of the one-dimensional convolutional neural network, realize real-time production control and optimization, and can realize real-time monitoring, intelligent adjustment and optimization control of the oilfield gathering and transportation system, thereby improving production efficiency, reducing energy consumption, and adapting to the complexity and variability of the oilfield production environment; and can effectively capture the characteristics and patterns in the oilfield production data, thereby providing more accurate prediction and decision-making for the oilfield gathering and transportation system.
[0070] The production operation status optimization method of the oilfield gathering and transportation system provided by the embodiment of the present invention can reduce the impact of adverse factors such as human errors and fatigue by using a one-dimensional convolutional neural network to achieve intelligent production control, and can also assist oilfield operators in discovering potential safety hazards, reduce the risk of production accidents, and improve the reliability of the production process; and it helps to reduce energy consumption, reduce production costs, reduce environmental pollution and waste of resources, and improve the sustainable development level of the oilfield.
[0071] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings.
[0072] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS
[0073] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:
[0074] Figure 1 A schematic diagram of the steps of a method for optimizing the production and operation status of an oilfield gathering and transportation system provided by an embodiment of the present invention;
[0075] Figure 2 A schematic diagram of the steps of a training method for a prediction model based on a one-dimensional convolutional neural network provided in an embodiment of the present invention;
[0076] Figure 3 A schematic diagram of the structure of a one-dimensional convolutional neural network model provided by an embodiment of the present invention;
[0077] Figure 4 A schematic diagram of the structure of a production and operation state optimization device for an oilfield gathering and transportation system provided by an embodiment of the present invention;
[0078] Figure 5 A schematic diagram of the structure of a training device for a prediction model based on a one-dimensional convolutional neural network provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0079] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.
[0080] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.
[0081] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.
[0082] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.
[0083] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.
[0084] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.
[0085] It should be understood that the size of the serial numbers of the steps in the following embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
[0086] In order to illustrate the technical solution of the present application, a specific embodiment is provided below for illustration.
[0087] The inventors found that in the prior art, the traditional oilfield gathering and transportation system mainly relies on manual monitoring and experience judgment to make operational adjustments. This method has the following problems:
[0088] Poor real-time performance: The oil field production environment is complex and changeable. Manual monitoring and judgment are difficult to respond to production changes in real time, which may lead to inefficiency and energy waste in the production process.
[0089] Lack of accuracy: Human experience has limitations, and it is difficult to make accurate judgments and decisions on complex oilfield production phenomena;
[0090] Low reliability: Manual operations are susceptible to factors such as fatigue and errors, which may lead to safety hazards and production accidents.
[0091] In order to solve the above problems, in recent years, the oil field production field has begun to try to introduce artificial intelligence technology, using data mining, pattern recognition and other methods to achieve intelligent production control and optimization. However, the current research and application of artificial intelligence for oil field gathering and transportation systems is still in its early stages, and there are mainly the following problems:
[0092] High model complexity: The oilfield gathering and transportation system involves multiple wellheads, pipelines, and processing facilities, which have complex mathematical relationships with each other. It is necessary to build a model with a certain degree of complexity to accurately reflect the production process;
[0093] Low computational efficiency: Real-time optimization of oilfield gathering and transportation systems requires rapid computational model outputs. Traditional artificial intelligence methods, such as support vector machines and neural networks, are insufficient in computational efficiency.
[0094] Based on this, the inventors have made the present invention after further research and development, providing a method, device, equipment and medium for optimizing the production and operation status of an oilfield gathering and transportation system.
[0095] Embodiment 1
[0096] The embodiment of the present invention provides a method for optimizing the production and operation status of an oilfield gathering and transportation system, referring to Figure 1 As shown, the method includes:
[0097] S101: Obtain at least one operating parameter within a preset historical time period, and use a pre-trained prediction model based on a one-dimensional convolutional neural network to perform prediction to obtain at least one corresponding predicted operating parameter result.
[0098] In this embodiment, the prediction model based on the one-dimensional convolutional neural network in the above step S101 needs to be pre-built and trained. Since the general one-dimensional convolutional neural network (1D-CNN) includes an input layer, a convolution layer, a pooling layer, a fully connected layer and an output layer, the inventor introduced the time series data feature in order to make the model better adapt to the characteristics of the oil field gathering and transportation system, and added multiple convolution layers and pooling layers to improve the expression ability of the model.
[0099] In this embodiment, after the above-mentioned one-dimensional convolutional neural network prediction model is established, the model needs to be trained. Figure 2 As shown, the prediction model is obtained by training in the following manner:
[0100] S201: dividing the historical operation data of the oilfield gathering and transportation system acquired in advance into a plurality of training sets and a plurality of test sets;
[0101] S202: Input each of the training sets into a pre-built one-dimensional convolutional neural network model for training;
[0102] S203: Input each of the test sets into the trained one-dimensional convolutional neural network model for testing to obtain a test loss value;
[0103] S204: Repeat the above model training and testing steps to obtain the one-dimensional convolutional neural network model with the smallest test loss value as the prediction model.
[0104] In the above method, before dividing the pre-acquired historical operation data of the oilfield gathering and transportation system into a training set and a test set, it also includes:
[0105] The historical operation data of the oilfield gathering and transportation system acquired in advance are preprocessed; the preprocessing method includes at least one of normalization, noise reduction and feature selection.
[0106] In the embodiment of the present invention, the historical operation data of the oilfield gathering and transportation system is scaled according to its value range so as to fall into the range of [0,1], so that the data can be at the same level, which is convenient for subsequent processing and analysis.
[0107] In the embodiment of the present invention, the historical operation data of the oilfield gathering and transportation system is smoothed by adopting the sliding average method, and the noise component in the data is reduced and the noise effect is eliminated by calculating the average value of the data within a certain window size.
[0108] In the embodiment of the present invention, the features most relevant to a parameter adjustment are selected based on correlation analysis and domain knowledge. For example, the features most relevant to pressure adjustment are temperature, flow, etc. This can eliminate the interference of irrelevant or redundant features on the model and improve the prediction performance of the trained model.
[0109] The method for optimizing the production and operation status of an oilfield gathering and transportation system provided by the embodiment of the present invention pre-processes data by means of normalization, noise reduction and feature selection, thereby effectively improving data quality.
[0110] In this embodiment, the above-mentioned historical operation data of the oilfield gathering and transportation system acquired in advance includes flow rate (Q, m 3 / d), pressure (P, psi), temperature (T, °C) and water content (W, %) and other historical operating data. In a specific embodiment, assume that there are 10 oil wells in an oil field, and each oil well has corresponding data acquisition equipment, including sensors such as pressure, temperature, flow rate and water content. The historical operating data of the flow rate, pressure, temperature and water content of the oil field gathering and transportation system are obtained, and then the obtained operating data are preprocessed by normalization, noise reduction and feature selection.
[0111] In a specific embodiment, the following data collected from an oil well is preprocessed as an example:
[0112] Pressure: 1000psi;
[0113] Temperature: 80℃;
[0114] Flow rate: 100m 3 / d;
[0115] Moisture content: 15%;
[0116] The technical means used for pretreatment are as follows:
[0117] (1) Normalization: Scale each data item according to its value range so that it falls within the range of [0,1]. This will keep the data at the same level, making it easier to process and analyze later.
[0118] For example, normalize pressure, temperature, flow rate, and water cut to the following ranges:
[0119] Pressure: 0-5000psi → 0-1;
[0120] Temperature: 0-100℃→0-1;
[0121] Flow rate: 0-1000m 3 / d→0-1;
[0122] Moisture content: 0-100% → 0-1;
[0123] The normalized data of oil well 1 is obtained as follows:
[0124] Pressure: 0.2;
[0125] Temperature: 0.8;
[0126] Flow rate: 0.1;
[0127] Moisture content: 0.15.
[0128] (2) Noise reduction: Use the sliding average method to smooth the data. By calculating the average value of the data within a certain window size, the noise component in the data can be reduced and the noise effect can be eliminated. For example, if the window size is set to 5, the sliding average of the pressure data of oil well 1 can be calculated.
[0129] (3) Feature selection: Based on correlation analysis and domain knowledge, select the features most relevant to a certain parameter adjustment. For example, the features most relevant to pressure adjustment are temperature, flow, etc. This can eliminate the interference of irrelevant or redundant features on the model and improve the prediction performance of the model.
[0130] After the previously acquired historical operation data of the oilfield gathering and transportation system is preprocessed in the above manner, the ratio of the training set to the test set is set, and the data is divided into multiple training sets and multiple test sets. For example, after the historical operation data of the oilfield gathering and transportation system in the past year is preprocessed, the data of the first 9 months of oil wells 1 to 8 are used as the training set, and the data of the last 3 months of oil wells 1 to 8 and the full year data of oil wells 9 to 10 are used as the test set.
[0131] In the above step S202, each of the training sets is input into a pre-built one-dimensional convolutional neural network model for training, including:
[0132] Input each of the training sets into a pre-built one-dimensional convolutional neural network model to obtain corresponding prediction training results;
[0133] Determine the loss value between each of the predicted training results and the corresponding true label result;
[0134] Determine the gradient of the model parameters corresponding to the loss value, and update the model parameters according to the gradient of the model parameters to obtain a trained one-dimensional convolutional neural network model.
[0135] In a specific embodiment, the gradient of the above model parameters includes a first-order moment estimation and a second-order moment estimation; according to the gradient of the model parameters, the model parameters are updated in the following manner:
[0136] Based on the following formula 7, use the Adam optimizer to update the parameters:
[0137] θ _t+1= θ _t - learning_rate * (m _t / (sqrt(v _t ) + ε)) Formula 7;
[0138] In the formula, θ _t is the model parameter; m _t is the first-order moment estimate; v _t is the second-order moment estimate; learning_rate is the learning rate; ε is a small constant to prevent division by zero.
[0139] Reference Figure 3 As shown, the number of convolutional layers and pooling layers is N, and the value of N is manually adjusted according to the accuracy of the model. The model structure parameters are preset, such as the size of the convolution kernel (filters), the size of the pooling window, and the number of neurons in the fully connected layer. Considering the scenario in which the present invention is applied, assuming that each oil well is equipped with corresponding data acquisition equipment, and the operating parameters collected include pressure, temperature, flow rate and water content, then a training model based on a one-dimensional convolutional neural network can be pre-built to perform model training to predict parameters such as pressure, temperature, flow rate and water content, so as to optimize the production and operation status of the oilfield gathering and transportation system. The detailed model construction process is as follows:
[0140] (1) Input layer: Receives preprocessed operating parameters, including pressure (P), temperature (T) and flow rate (Q), etc., recorded as input features. Other operating parameters may also be included, such as water content (W). In a specific embodiment, assume that the dimension of the input data is (1,4), where 4 represents the four features we selected, namely, normalized pressure, temperature, flow rate and water content. Taking oil well 1 as an example, the input preprocessed operating data is (0.2, 0.8, 0.1, 0.15).
[0141] (2) Convolution layer: Use several one-dimensional convolution kernels to extract local features of the input data. Assuming that 8 convolution kernels are used and the size of each convolution kernel is 3, that is, kernel_size = 3, the output dimension of the convolution layer is (1,8). The calculation formula of the convolution layer is:
[0142] Y[i] = f(sum(X[i:i+2] * W[i]) + b) Formula 8;
[0143] In the formula, Y is the output of the convolution layer; X is the data of the input layer; W is the convolution kernel weight; b is the bias term; f is the activation function, such as ReLU; i is the sliding index of the convolution kernel on the input data.
[0144] In the above formula 8, the convolution kernel weight (W) is gradually adjusted during the model training process through optimization algorithms such as gradient descent. At the beginning of training, the convolution kernel weight is usually initialized using random values or a specific initialization strategy, such as Xavier initialization or He initialization. The weight value is updated according to the back-propagation algorithm and the gradient of the loss function.
[0145] The bias term (b) is adjusted by the optimization algorithm during the model training process. The initial value of the bias term is usually set to 0 or a small value close to 0.
[0146] The sliding index (i) is determined by the stride of the convolution operation and the size of the input data. For input data of length 10, a convolution kernel of size 3, and a convolution operation of stride 1, the sliding index i ranges from 0 to 7 (including 0 and 7).
[0147] The method for optimizing the production and operation status of an oilfield gathering and transportation system provided in an embodiment of the present invention extracts local features through the convolution layer of a one-dimensional convolutional neural network, thereby reducing the complexity of the model to a certain extent; moreover, it can reduce the computing resources and time for processing complex problems and improve computing efficiency.
[0148] (3) Pooling layer: Downsample the output of the convolutional layer to reduce the amount of computation and extract key features. For example, the max pooling method can be used with a pooling window size of 2, i.e., pool_size = 2, and the output dimension of the pooling layer is (1, 4).
[0149] (4) Fully connected layer: Connect the output of the pooling layer to the neuron nodes to achieve nonlinear mapping. For example, the number of nodes in the fully connected layer can be set to 16, and the ReLU activation function can be used. The calculation formula of the fully connected layer is:
[0150] Z = f(W * Y + b) Formula 9;
[0151] Where Z is the output of the fully connected layer; Y is the output of the pooling layer; W is the weight matrix; b is the bias vector; and f is the activation function, such as the ReLU activation function.
[0152] In the above formula 9, the weight matrix (W) is gradually adjusted by optimization algorithms such as gradient descent during the model training process; the bias vector (b) is automatically adjusted by the optimization algorithm during the model training process.
[0153] (5) Output layer: outputs the predicted operating parameter results, such as the predicted pressure results. For example, the number of nodes in the output layer can be set to 1, and a linear activation function can be used. The linear activation function plays a role in directly outputting the predicted operating parameter results in the output layer. The linear activation function is usually defined as f(x) = x, that is, the output value is equal to the input value, and the input of the linear activation function is the output of the fully connected layer. The calculation formula of the output layer is:
[0154] P = W * Z + b Formula 10;
[0155] Where P is the predicted operating parameter result of the output layer; Z is the output of the fully connected layer; W is the weight matrix; and b is the bias vector.
[0156] In the above formula 10, the weight matrix (W) is gradually adjusted by an optimization algorithm such as the gradient descent method during the model training process; the bias vector (b) is automatically adjusted by the optimization algorithm during the model training process.
[0157] In this embodiment, the training set is input into a pre-built one-dimensional convolutional neural network model for training. The specific process of training is as follows:
[0158] (1) Setting training parameters
[0159] Set hyperparameters such as training rounds (epochs), batch size (batch_size) and learning rate (learning_rate). For example, set the values as follows:
[0160] epochs = 100;
[0161] batch_size = 32;
[0162] learning_rate=0.001.
[0163] (2) Data batching
[0164] The training data is divided into batches according to the batch size, and the training set is divided into multiple subsets, each of which contains samples of the batch size. A similar method is used to batch the test set into multiple test subsets for testing.
[0165] (3) Model training
[0166] In each training epoch, all batches of training data are traversed. For each training subset, the following steps are performed:
[0167] a. Input the subset data into the model and calculate the output results;
[0168] b. Calculate the loss value based on the predicted results and the true label, using the mean square error (MSE);
[0169] c. Calculate the gradient of the loss value to the model parameters using the backpropagation algorithm (BP);
[0170] d. Update the model parameters and use the Adam optimizer to update the parameters according to the following formula:
[0171] θ _t +1 = θ _t - learning_rate * (m _t / (sqrt(v _t ) + ε)) Formula 7;
[0172] In the formula, θ _t is the model parameter; m _t is the first-order moment estimate; v _t is the second-order moment estimate; learning_rate is the learning rate; ε is a small constant to prevent division by zero.
[0173] In the above formula 7, m _t and v _t It is solved by iteration, setting the initial value to 0, and then iteratively updating it by the formula:
[0174] m _t = β 1 * m _t-1 + (1 - β 1 ) * gradient formula 11;
[0175] Where gradient is the gradient of the model parameters; β 1 Usually set to 0.9;
[0176] v _t = β 2 * v _t-1 + (1 - β 2 ) * gradient^2 formula 12;
[0177] Where gradient is the gradient of the model parameters; β 2 Usually set to 0.999;
[0178] The specific implementation of the above iteration can refer to the detailed description in the prior art and is not specifically limited here.
[0179] In the above formula 7, ε is a small constant to prevent division by zero. Usually, the value range of constant ε is 1e -8 To 1e -10 Between. With constant ε = 1e -8 For example, assuming that the first-order moment of the current gradient is estimated to be m _t =0.1, the second-order moment is estimated to be v _t =0, then the updated parameter change is:
[0180] θ _t+1 =θ _t -learning_rate*(m _t / (sqrt(v _t )+ε))
[0181] =θ _t -learning_rate*(0.1 / (sqrt(0)+1e -8 ))
[0182] =θ _t -1e -7 *0.1
[0183] It can be seen that the value of constant ε has very little effect on parameter update and is usually set at 1e -8 To 1e -10 In between.
[0184] According to the training method of steps (1) to (3) above, all training sets are input into the model for training. After each training round, the training loss value of the model on the training set is summarized and the average loss value of the training is calculated. For example, in the first round of training, the training loss value is 0.05; in the 50th round of training, the training loss value is reduced to 0.01; and finally in the 100th round of training, the training loss value stabilizes at 0.001.
[0185] After each training round, the model needs to be tested using the test set. Input the batched test subsets into the model and calculate the output results. Calculate the loss value on the test set based on the predicted results and the true labels, that is, the test loss value. After each verification round, the test loss values of the model are summarized and the average loss value of the test is calculated. The specific implementation method of the above test can refer to the detailed description in the prior art and is not specifically limited here.
[0186] Record the training loss and test loss of the model in each training round. Evaluate the model performance by observing the changing trends of the training loss and test loss. If the training loss of the model decreases while the test set loss increases, it means that the model is overfitting. At this point, you can consider taking measures to make adjustments, such as increasing training data, reducing model complexity, and adding regularization terms.
[0187] During the training process, the prediction error can be obtained according to the test loss value, and the model parameters with the smallest prediction error are used as the optimal model. In this way, in practical applications, the saved optimal model parameters can be directly loaded for prediction and adjustment. The optimal model is used as the prediction model based on the one-dimensional convolutional neural network in this embodiment. In a specific embodiment, it is assumed that the prediction errors of the model in oil wells 9 and 10 are 0.002 and 0.003 respectively, that is, the total prediction error is 0.005. By comparing the training results of different rounds, it is found that the prediction error of the model on the test set is the smallest in the 80th round, and the model parameters in the 80th round are selected as the optimal model, that is, the prediction model in this embodiment.
[0188] In this embodiment, a prediction model based on a one-dimensional convolutional neural network is used to monitor the oilfield gathering and transportation system in real time. The operating data collected in real time by the oilfield gathering and transportation system, or the historical operating data of the oilfield gathering and transportation system within a preset historical time period, is input into the model for prediction, such as the historical operating data of the oilfield gathering and transportation system in the past year. The historical operating data of the oilfield gathering and transportation system includes at least one operating parameter, such as oil well production status, pipeline pressure, temperature, and water content. In this embodiment, the pump station pressure, temperature, and flow parameters are taken as an example, and the pump station pressure, temperature, and flow parameters of the oilfield gathering and transportation system within the preset historical time period that are collected are input into the prediction model based on the one-dimensional convolutional neural network for prediction, and the corresponding prediction parameter results are obtained. The corresponding prediction parameter results are then used to intelligently adjust the operating status of the oilfield gathering and transportation system, adjust the pump station pressure, temperature, and flow parameters, and optimize the operating efficiency of the oilfield gathering and transportation system.
[0189] S102: Determine an operating parameter error between each of the predicted operating parameter results and a corresponding preset target operating parameter result.
[0190] In a specific embodiment, in the above step S102, the predicted operating parameter results include predicted pump station pressure results, predicted temperature results and predicted flow results; the preset target operating parameter results include preset target pump station pressure, preset target temperature and preset target flow; the operating parameter errors include pump station pressure error, temperature error and flow error;
[0191] The above-mentioned determination of the operating parameter error between each of the predicted operating parameter results and the corresponding preset target operating parameter result specifically includes:
[0192] Based on the following formula 1, the pump station pressure error is calculated:
[0193] ΔP = P _target -P _pred Formula 1;
[0194] Where ΔP is the pressure error of the pump station; P _target is the preset target pump station pressure; P _pred The predicted pump station pressure result;
[0195] Based on the following formula 2, the temperature error is calculated:
[0196] ΔT = T _target -T _pred Formula 2;
[0197] Where ΔT is the temperature error; T _target is the preset target temperature; T _pred is the predicted temperature result;
[0198] Based on the following formula 3, the flow error is calculated:
[0199] ΔQ = Q _target -Q _pred Formula 3;
[0200] Where ΔQ is the flow error; Q _target is the preset target flow rate; Q _pred This is the predicted traffic result.
[0201] In the above step S102, it is necessary to preset the target operating parameters corresponding to each parameter to prepare for the intelligent adjustment of the operating status of the oilfield gathering and transportation system. The preset target operating parameters are the operating parameters of the oilfield gathering and transportation system that are pre-set to meet the expected production status. The error between each preset target operating parameter and the corresponding prediction result is calculated, and then the operating parameters of the oilfield gathering and transportation system are intelligently adjusted according to the values of these errors.
[0202] S103: Determine at least one optimized operating parameter of the oilfield gathering and transportation system according to each obtained operating parameter error and the corresponding current operating parameter, so as to dynamically adjust the real-time operating state of the oilfield gathering and transportation system.
[0203] In the above step S103, the current operating parameters include the current pump station pressure, current temperature and current flow rate; the optimized operating parameter results include the optimized pump station pressure, optimized temperature and optimized flow rate;
[0204] In a specific embodiment, the above-mentioned determining at least one optimized operating parameter of the oilfield gathering and transportation system based on each obtained operating parameter error and the corresponding current operating parameter to dynamically adjust the real-time operating state of the oilfield gathering and transportation system specifically includes:
[0205] Based on the following formula 4, the optimized pump station pressure is calculated:
[0206] P _new = P _current + k1 * ΔP Formula 4;
[0207] Where P _new is the pump station pressure after optimization; P _current is the current pump station pressure; k1 is the first preset adjustment coefficient; ΔP is the pump station pressure error;
[0208] Based on the following formula 5, the optimized temperature is calculated:
[0209] T _new = T _current + k2 * ΔT Formula 5;
[0210] Where, T _new is the optimized temperature; T _current is the current temperature; k2 is the second preset adjustment coefficient; ΔT is the temperature error;
[0211] Based on the following formula 6, the optimized flow is calculated:
[0212] Q _new = Q _current + k3 * ΔQ Formula 6;
[0213] In the formula, Q _new is the optimized flow rate; Q _current is the current flow rate; k3 is the third preset adjustment coefficient; ΔQ is the flow rate error;
[0214] The real-time operation status of the oilfield gathering and transportation system is dynamically adjusted according to the optimized pump station pressure, optimized temperature and optimized flow.
[0215] In the above step S103, k1, k2 and k3 are all preset adjustment coefficients, which are used to control the amplitude of adjustment and can be adjusted according to actual production conditions and manual experience. The specific parameter sizes can be determined according to actual needs and are not specifically limited here.
[0216] In the process of real-time monitoring of the oilfield gathering and transportation system by the prediction model based on the one-dimensional convolutional neural network, the prediction model based on the one-dimensional convolutional neural network can be dynamically adjusted according to the real-time operation status of the oilfield gathering and transportation system. The dynamic adjustment method is as follows:
[0217] Acquire at least one real-time operating parameter collected during the operation of the oilfield gathering and transportation system within a preset operating time period;
[0218] Retraining the prediction model according to at least one real-time operating parameter within the preset operating time period, and updating the model parameters using a stochastic gradient descent algorithm to obtain an updated prediction model;
[0219] The updated prediction model is used to perform prediction to obtain at least one corresponding prediction operation parameter result.
[0220] The production and operation status optimization method of the oilfield gathering and transportation system provided by the embodiment of the present invention realizes flexible adjustment and optimization according to changes in the oilfield production environment by dynamically adjusting the one-dimensional convolutional neural network prediction model, thereby enhancing generalization ability and scalability; and can realize adaptive optimization, and when the oilfield production environment changes, it can still maintain high prediction accuracy and control effect, adapt to the complexity and variability of the oilfield production environment, and has high practical application value and broad development prospects.
[0221] In the embodiment of the present invention, when the oil field production environment changes, such as oil field production fluctuations and pipeline aging, the model is fine-tuned by adding training data, adjusting the model structure, and updating model parameters to adapt to the complexity and variability of the oil field production environment. The specific implementation is as follows:
[0222] (1) Add training data: Based on the monitored changes in the production environment, collect data related to the new environment and add it to the original training set. The main body of monitoring here is the manual operation and maintenance personnel. For example, if a new problem occurs at a wellhead, the operation and maintenance personnel can analyze the sensor data, monitor the environmental changes, and collect relevant data in a timely manner to update the model and optimize the control strategy.
[0223] (2) Adjust the model structure: Adjust the model structure according to the characteristics of the new environment. By adding or reducing convolutional layers and pooling layers, the model's expressiveness and adaptability can be improved.
[0224] (3) Update model parameters: Retrain the model using the increased training data set. Use the Stochastic Gradient Descent (SGD) algorithm to update the model parameters. During the training process, set appropriate hyperparameters such as learning rate, batch size, and number of training rounds to control the update speed and overfitting degree of the model. The retraining process here is different from the initial training. The batch gradient descent (BGD) algorithm is usually used during initial training, that is, the gradient of the loss function is calculated on the entire training set, and the gradient is used to update the model parameters. When retraining the model after updating the model parameters, different optimization algorithms can be used, such as the stochastic gradient descent algorithm and the adaptive moment estimation (Adam). The stochastic gradient descent algorithm is a stochastic optimization algorithm that randomly selects only one training sample each time to calculate the gradient and update the parameters. Compared with the batch gradient descent algorithm, it is more efficient.
[0225] Through the above steps S101 to 103, the production and operation status of the oilfield gathering and transportation system can be optimized, the production efficiency of the oilfield gathering and transportation system can be improved, and energy consumption can be reduced. During the implementation process, the operation data of the oilfield gathering and transportation system can be monitored and recorded in real time to provide data support for subsequent optimization and maintenance. At the same time, the model performance can be evaluated regularly to ensure the accuracy of model prediction and the effectiveness of the adjustment strategy to achieve economic, environmental and sustainable development. In addition, in an embodiment of the present invention, data preprocessing, model construction and training, and verification of the prediction model can cooperate with each other. During the maintenance and upgrade process, the processing processes at different stages can be optimized and improved in a targeted manner to ensure the stable operation of the optimized production and operation status of the oilfield gathering and transportation system.
[0226] Embodiment 2
[0227] Based on the same inventive concept, the embodiment of the present invention also provides a training method for a prediction model based on a one-dimensional convolutional neural network, referring to Figure 2 As shown, the method includes:
[0228] S201: dividing the historical operation data of the oilfield gathering and transportation system acquired in advance into a plurality of training sets and a plurality of test sets;
[0229] S202: Input each of the training sets into a pre-built one-dimensional convolutional neural network model for training;
[0230] S203: Input each of the test sets into the trained one-dimensional convolutional neural network model for testing to obtain a test loss value;
[0231] S204: Repeat the above model training and testing steps to obtain the one-dimensional convolutional neural network model with the smallest test loss value as the prediction model.
[0232] The specific implementation of the above steps S201 to S204 is shown in Example 1 and will not be described in detail here.
[0233] Embodiment 3
[0234] Based on the same inventive concept, the embodiment of the present invention also provides a production operation state optimization device for an oilfield gathering and transportation system, referring to Figure 4 As shown, the device comprises:
[0235] The prediction parameter module 101 is used to obtain at least one operating parameter within a preset historical time period, and use a pre-trained prediction model based on a one-dimensional convolutional neural network to perform prediction to obtain at least one corresponding predicted operating parameter result;
[0236] An error determination module 102 is used to determine an operating parameter error between each of the predicted operating parameter results and a corresponding preset target operating parameter result;
[0237] The parameter optimization module 103 is used to determine at least one optimized operating parameter of the oilfield gathering and transportation system according to each obtained operating parameter error and the corresponding current operating parameter, so as to dynamically adjust the real-time operating state of the oilfield gathering and transportation system.
[0238] Embodiment 4
[0239] Based on the same inventive concept, the embodiment of the present invention also provides a training device for a prediction model based on a one-dimensional convolutional neural network, referring to Figure 5 As shown, the device comprises:
[0240] The first training module 201 is used to divide the pre-acquired historical operation data of the oilfield gathering and transportation system into multiple training sets and multiple test sets;
[0241] A second training module 202, configured to input each of the training sets into a pre-built one-dimensional convolutional neural network model for training;
[0242] The third training module 203 is used to input each of the test sets into the trained one-dimensional convolutional neural network model for testing to obtain a test loss value;
[0243] The fourth training module 204 is used to repeat the above model training and testing steps to obtain the one-dimensional convolutional neural network model with the smallest test loss value as the prediction model.
[0244] Embodiment 5
[0245] Based on the same inventive concept, an embodiment of the present invention also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, it implements the production operation status optimization method of the oilfield gathering and transportation system as described in the above embodiment one and / or the training method of the prediction model based on the one-dimensional convolutional neural network as described in the above embodiment two.
[0246] Embodiment 6
[0247] Based on the same inventive concept, an embodiment of the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the production operation status optimization method of the oilfield gathering and transportation system as described in the above-mentioned embodiment 1 and / or the training method of the prediction model based on the one-dimensional convolutional neural network as described in the above-mentioned embodiment 2.
[0248] Embodiment 7
[0249] Based on the same inventive concept, an embodiment of the present invention also provides a computer program product comprising instructions. When the computer program product is run on a computer device, the computer device executes the production operation status optimization method of the oilfield gathering and transportation system described in the above embodiment one and / or the training method of the prediction model based on the one-dimensional convolutional neural network described in the above embodiment two.
[0250] Embodiment 8
[0251] Based on the same inventive concept, an embodiment of the present invention also provides a chip, the chip includes a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run a computer program or instruction to implement the production operation status optimization method of the oilfield gathering and transportation system as described in the above embodiment one and / or the training method of the prediction model based on the one-dimensional convolutional neural network as described in the above embodiment two.
[0252] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention 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 and optical storage, etc.) containing computer-usable program code.
[0253] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks 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 processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device 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.
[0254] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising 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.
[0255] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.
[0256] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.
Claims
1. A method for optimizing the production and operation status of an oilfield gathering and transportation system, characterized in that: include: Obtain at least one operating parameter within a preset historical time period, and use a pre-trained one-dimensional convolutional neural network-based prediction model to perform predictions to obtain at least one corresponding predicted operating parameter result; Determining an operating parameter error between each of the predicted operating parameter results and a corresponding preset target operating parameter result; At least one optimized operating parameter of the oilfield gathering and transportation system is determined according to each obtained operating parameter error and the corresponding current operating parameter, so as to dynamically adjust the real-time operating status of the oilfield gathering and transportation system.
2. The method according to claim 1, characterized in that The predicted operating parameter results include predicted pump station pressure results, predicted temperature results and predicted flow results; the preset target operating parameter results include preset target pump station pressure, preset target temperature and preset target flow; The operating parameter errors include pump station pressure error, temperature error and flow error; The determining of the operating parameter error between each of the predicted operating parameter results and the corresponding preset target operating parameter result includes: Based on the following formula 1, the pump station pressure error is calculated: ΔP = P _target -P _pred Formula 1; Where ΔP is the pressure error of the pump station; P _target is the preset target pump station pressure; P _pred The predicted pump station pressure result; Based on the following formula 2, the temperature error is calculated: ΔT = T _target -T _pred Formula 2; Where ΔT is the temperature error; T _target is the preset target temperature; T _pred is the predicted temperature result; Based on the following formula 3, the flow error is calculated: ΔQ = Q _target -Q _pred Formula 3; Where ΔQ is the flow error; Q _target is the preset target flow rate; Q _pred This is the predicted traffic result.
3. The method according to claim 1, characterized in that The current operating parameters include the current pump station pressure, current temperature and current flow rate; the optimized operating parameter results include the optimized pump station pressure, optimized temperature and optimized flow rate; Determining at least one optimized operating parameter of the oilfield gathering and transportation system according to each obtained operating parameter error and the corresponding current operating parameter to dynamically adjust the real-time operating state of the oilfield gathering and transportation system includes: Based on the following formula 4, the optimized pump station pressure is calculated: P _new = P _current + k1 * ΔP Formula 4; Where P _new is the pump station pressure after optimization; P _current is the current pump station pressure; k1 is the first preset adjustment coefficient; ΔP is the pump station pressure error; Based on the following formula 5, the optimized temperature is calculated: T _new = T _current + k2 * ΔT Formula 5; Where, T _new is the optimized temperature; T _current is the current temperature; k2 is the second preset adjustment coefficient; ΔT is the temperature error; Based on the following formula 6, the optimized flow is calculated: Q _new = Q _current + k3 * ΔQ Formula 6; In the formula, Q _new is the optimized flow rate; Q _current is the current flow rate; k3 is the third preset adjustment coefficient; ΔQ is the flow rate error; The real-time operation status of the oilfield gathering and transportation system is dynamically adjusted according to the optimized pump station pressure, optimized temperature and optimized flow.
4. The method according to claim 1, characterized in that The prediction model is obtained by training in the following way: Dividing the previously acquired historical operation data of the oilfield gathering and transportation system into multiple training sets and multiple test sets; Input each of the training sets into a pre-built one-dimensional convolutional neural network model for training; Input each of the test sets into the trained one-dimensional convolutional neural network model for testing to obtain a test loss value; Repeat the above model training and testing steps to obtain the one-dimensional convolutional neural network model with the smallest test loss value as the prediction model.
5. The method according to claim 4, characterized in that Inputting each of the training sets into a pre-built one-dimensional convolutional neural network model for training includes: Input each of the training sets into a pre-built one-dimensional convolutional neural network model to obtain corresponding prediction training results; Determine the loss value between each of the predicted training results and the corresponding true label result; Determine the gradient of the model parameters corresponding to the loss value, and update the model parameters according to the gradient of the model parameters to obtain a trained one-dimensional convolutional neural network model.
6. The method according to claim 5, characterized in that The gradient of the model parameters includes a first-order moment estimate and a second-order moment estimate; according to the gradient of the model parameters, the model parameters are updated in the following manner: Based on the following formula 7, use the Adam optimizer to update the parameters: θ _t+1 = θ _t - learning_rate * (m _t / (sqrt(v _t ) + ε)) Formula 7; In the formula, θ _t is the model parameter; m _t is the first-order moment estimate; v _t is the second-order moment estimate; learning_rate is the learning rate; ε is a small constant to prevent division by zero.
7. The method according to claim 4, characterized in that Before dividing the previously acquired historical operation data of the oilfield gathering and transportation system into a training set and a test set, it also includes: The historical operation data of the oilfield gathering and transportation system acquired in advance are preprocessed; the preprocessing method includes at least one of normalization, noise reduction and feature selection.
8. The method according to claim 1, characterized in that Also includes: Acquire at least one real-time operating parameter collected during the operation of the oilfield gathering and transportation system within a preset operating time period; Retraining the prediction model according to at least one real-time operating parameter within the preset operating time period, and updating the model parameters using a stochastic gradient descent algorithm to obtain an updated prediction model; The updated prediction model is used to perform prediction to obtain at least one corresponding prediction operation parameter result.
9. A training method for a prediction model based on a one-dimensional convolutional neural network, characterized in that: include: Dividing the previously acquired historical operation data of the oilfield gathering and transportation system into multiple training sets and multiple test sets; Input each of the training sets into a pre-built one-dimensional convolutional neural network model for training; Input each of the test sets into the trained one-dimensional convolutional neural network model for testing to obtain a test loss value; Repeat the above model training and testing steps to obtain the one-dimensional convolutional neural network model with the smallest test loss value as the prediction model.
10. A production operation status optimization device for an oilfield gathering and transportation system, characterized in that: include: A prediction parameter module is used to obtain at least one operating parameter within a preset historical time period, and use a pre-trained prediction model based on a one-dimensional convolutional neural network to make a prediction to obtain at least one corresponding predicted operating parameter result; An error determination module is used to determine an operating parameter error between each of the predicted operating parameter results and a corresponding preset target operating parameter result; The parameter optimization module is used to determine at least one optimized operating parameter of the oilfield gathering and transportation system according to each obtained operating parameter error and the corresponding current operating parameter, so as to dynamically adjust the real-time operating state of the oilfield gathering and transportation system.
11. A training device for a prediction model based on a one-dimensional convolutional neural network, characterized in that: include: The first training module is used to divide the pre-acquired historical operation data of the oilfield gathering and transportation system into multiple training sets and multiple test sets; A second training module is used to input each of the training sets into a pre-built one-dimensional convolutional neural network model for training; A third training module is used to input each of the test sets into the trained one-dimensional convolutional neural network model for testing to obtain a test loss value; The fourth training module is used to repeat the above model training and testing steps to obtain the one-dimensional convolutional neural network model with the smallest test loss value as the prediction model.
12. A computer-readable storage medium, in which instructions are stored. When the instructions are executed on a terminal, the terminal executes the production operation status optimization method of the oil field gathering and transportation system as described in any one of claims 1 to 8 and / or the training method of the prediction model based on the one-dimensional convolutional neural network as described in claim 9.
13. A computer device, characterized in that: It includes a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the production operation status optimization method of the oilfield gathering and transportation system as described in any one of claims 1 to 8 and / or the training method of the prediction model based on the one-dimensional convolutional neural network as described in claim 9.
14. A computer program product comprising instructions, which, when executed on a computer device, enables the computer device to execute the production operation status optimization method of an oilfield gathering and transportation system as described in any one of claims 1 to 8 and / or the training method of a prediction model based on a one-dimensional convolutional neural network as described in claim 9.
15. A chip, comprising a processor and a communication interface, wherein the communication interface and the processor are coupled, and the processor is used to run a computer program or instruction to implement the production operation status optimization method of an oilfield gathering and transportation system as described in any one of claims 1 to 8 and / or the training method of a prediction model based on a one-dimensional convolutional neural network as described in claim 9.
Citation Information
Cited By
Modularization-based oil field gathering and transportation intelligent operation and maintenance system and method
CN119849707A
Modular-based Intelligent Operation and Maintenance System and Method for Oilfield Gathering and Transportation
CN119849707B
