Method for predicting carbon emission of offshore oil and gas field

Abstract

The invention discloses a carbon emission prediction method for an offshore oil and gas field, and the method comprises the steps: obtaining original data of the offshore oil and gas field, including productivity data, energy consumption data and oil and gas field types; preprocessing the obtained original data to obtain preprocessed data; on the basis of the preprocessed data, calculating the carbon emissions of a plurality of production links; according to a calculation result of the carbon emission, screening characteristic quantities by adopting a Pearson correlation analysis method; inputting the screened characteristic quantity into a BP neural network model, wherein the BP neural network model outputs a predicted value of carbon emission; and acquiring carbon satellite remote sensing data, and calibrating the predicted value of the carbon emission of the BP neural network model through a multi-scale fusion algorithm to obtain a calibrated predicted value of the carbon emission. According to the method and the system, the multi-source information such as the carbon satellite remote sensing data, the offshore oil and gas field data and the like is fused, and the multi-scale fusion algorithm is combined, so that high-precision and real-time dynamic prediction from the total area quantity to the single-platform emission is realized.

Description

Technical Field

[0001] The present application relates to the field of carbon dioxide monitoring, and in particular to a method for predicting carbon emissions from offshore oil and gas fields. Background Art

[0002] With the continued development of the global economy and the growing population, energy demand is increasing. Because offshore oil and gas production is located far from shore, energy supply and consumption management face unique challenges. Therefore, in-depth analysis of oil and gas platform energy consumption is particularly important.

[0003] However, offshore oil and gas fields involve multiple production processes and emit a wide range of carbon emissions from numerous sources. Analyzing and predicting carbon emissions from offshore oil and gas fields is a critical issue urgently needed in the offshore oil and gas industry. It is also a crucial step toward achieving low-carbon development of offshore oil and gas fields and a crucial step in the industry's transition to sustainable development. With the continuous advancement of science and technology, offshore oil and gas field exploration, development, and production technologies are constantly being upgraded. However, traditional technical methods and production models are often associated with high carbon emissions. Therefore, reducing carbon emissions from offshore oil and gas fields through technological innovation and industrial upgrading has become an urgent need for the industry's development. Summary of the Invention

[0004] To solve the above problems, this application provides a carbon emission prediction method for offshore oil and gas fields, aiming to solve the problems of insufficient accuracy, poor dynamic adaptability and lack of multi-source data fusion in the existing technology of emission factor method and material balance method when monitoring carbon emissions from offshore oil and gas fields.

[0005] A first aspect of an embodiment of the present invention provides a method for predicting carbon emissions from an offshore oil and gas field, comprising:

[0006] S1. Obtaining raw data of offshore oil and gas fields, including production capacity data, energy consumption data, and oil and gas field types;

[0007] S2. Preprocessing the acquired raw data to obtain preprocessed data;

[0008] S3. Calculate the carbon emissions of multiple production links based on the pre-processed data;

[0009] S4. Based on the calculation results of carbon emissions, the characteristic quantities are screened by using the Pearson correlation analysis method;

[0010] S5. Inputting the selected feature values into a BP neural network model, wherein the BP neural network model outputs a predicted value of carbon emissions;

[0011] S6. Obtain carbon satellite remote sensing data, and calibrate the predicted value of carbon emissions of the BP neural network model using a multi-scale fusion algorithm to obtain a calibrated predicted value of carbon emissions.

[0012] In an optional embodiment, the production capacity data includes natural gas production, crude oil production, water production and water injection volume;

[0013] The energy consumption data include crude oil consumption, diesel consumption, natural gas consumption, flare gas emissions and vent air emissions;

[0014] The types of offshore oil and gas fields include conventional oil fields, conventional gas fields, heavy oil thermal recovery oil fields, low permeability oil fields, low permeability gas fields and high carbon gas fields.

[0015] In an optional embodiment, the preprocessing includes missing value processing, outlier processing, data type conversion and deduplication processing.

[0016] In an optional embodiment, the carbon emissions of the multiple production links include flare gas carbon emissions, vent air carbon emissions, crude oil combustion carbon emissions, diesel combustion carbon emissions and natural gas combustion carbon emissions.

[0017] In an optional implementation, the BP neural network model is iteratively optimized using a gradient descent algorithm, wherein the hidden layer activation function adopts a tangent S-type transfer function, and the output layer activation function adopts a linear transfer function.

[0018] In an optional embodiment, the carbon satellite remote sensing data obtained includes the carbon dioxide concentration in the area where the oil and gas field is located;

[0019] Calibrate carbon emission predictions based on regional CO2 concentrations:

[0020]

[0021] Where E is the regional carbon emissions, ΔC is the CO2 concentration difference inside and outside the region, h is the height of the atmospheric mixing layer, ρ is the air density, A is the regional area, and Δt is the time interval.

[0022] In an optional embodiment, the multi-scale fusion algorithm includes: hierarchically fusing the global emission constraints provided by carbon satellite data with the local prediction values of the BP neural network model.

[0023] In an optional embodiment, a Pearson correlation coefficient analysis method is used to screen characteristic quantities related to carbon emissions, and characteristic quantities with an absolute value of a correlation coefficient greater than 0.7 are screened.

[0024] In an optional embodiment, a BP neural network model is constructed, the model is trained using a training set, and the model performance is verified using a test set. The ratio of the training set to the test set is 4:1 or 9:1, and the model performance evaluation indicators include R2, MAE and RMSE.

[0025] In the disclosed embodiments, this application integrates multi-source information such as carbon satellite remote sensing data, platform sensor monitoring data, and meteorological data, and combines multi-scale fusion algorithms and dynamic assimilation algorithms to achieve high-precision, real-time dynamic prediction of emissions from regional totals to single-platform emissions.

[0026] This application can effectively adapt to the dynamic changes in carbon emissions caused by increased water content and equipment aging during the development of offshore oil and gas fields, break through the limitations of traditional methods in monitoring local emission hotspots, significantly improve the temporal and spatial resolution and accuracy of carbon emission predictions, and provide reliable technical support for low-carbon operation and carbon supervision of offshore oil and gas fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0028] Figure 1 This is a flow chart of a carbon emission prediction method for offshore oil and gas fields proposed in one embodiment of the present application;

[0029] Figure 2 This is a schematic diagram of a carbon emission prediction method for an offshore oil and gas field proposed in one embodiment of the present application;

[0030] Figure 3 This is the carbon emission prediction model training process proposed in one embodiment of the present application. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0032] Please refer to Figure 1 and Figure 2 , Figure 1 This is a flow chart of a method for predicting carbon emissions from offshore oil and gas fields proposed in one embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of a carbon emission prediction method for an offshore oil and gas field proposed in one embodiment of the present application. Figure 1 and Figure 2 As shown, a method for predicting carbon emissions from offshore oil and gas fields includes:

[0033] S1. Obtaining raw data of offshore oil and gas fields, including production capacity data, energy consumption data, and oil and gas field types;

[0034] In this example, the raw data sources include production databases, real-time sensor monitoring records, or public industry reports. The oil and gas field types must be clearly labeled (e.g., conventional oil fields, heavy oil thermal recovery fields, etc.) to facilitate subsequent analysis of the differential impacts of different offshore oil and gas field types on carbon emissions.

[0035] S2. Preprocessing the acquired raw data to obtain preprocessed data;

[0036] S3. Calculate the carbon emissions of multiple production links based on the pre-processed data;

[0037] S4. Based on the calculation results of carbon emissions, the characteristic quantities are screened by using the Pearson correlation analysis method;

[0038] In this embodiment, the linear relationship is measured by the Pearson correlation coefficient, and the correlation coefficient matrix is visualized with the help of a heat map to screen out the key driving factors. At this stage, it is necessary to combine domain knowledge to eliminate false correlations. For example, an increase in water injection volume may lead to a simultaneous increase in water production, but the main cause of actual carbon emissions may be the energy consumption of the water injection pump.

[0039] Correlation analysis was performed using Origin, which supports a variety of 2D / 3D graphs. The original data table was imported into Origin, and correlation analysis was performed using the Correlation Plot tool, a built-in correlation analysis plug-in.

[0040] S5. Inputting the selected feature values into a BP neural network model, wherein the BP neural network model outputs a predicted value of carbon emissions;

[0041] In this embodiment, the predicted value of the BP neural network model is compared with the true value to calculate the average relative error. If the error does not meet the accuracy requirement and is within the set number of iterations, the model parameters are adjusted and the BP neural network model is retrained until the training result meets the accuracy requirement or the number of training times reaches the number of iterations.

[0042] After the model training is completed, the sample data of the test set is input into the BP neural network model, and the predicted value is output. By comparing the predicted value of the BP neural network model with the true value, the average relative error is calculated. If the error requirement is not met, the model parameters are adjusted and the BP neural network model is retrained until the model test meets the error requirement. At this time, the corresponding BP neural network model is the offshore oil and gas field carbon emissions prediction model.

[0043] S6. Obtain carbon satellite remote sensing data, and calibrate the predicted value of carbon emissions of the BP neural network model using a multi-scale fusion algorithm to obtain a calibrated predicted value of carbon emissions.

[0044] In this embodiment, the carbon satellite data acquisition method may access data from the domestically produced TANSAT carbon satellite. Since the entire life cycle of an offshore oil and gas field can be divided into a construction period and a stable production period, when an oil and gas field is in the stable production period, its carbon emissions are relatively stable. Therefore, current carbon data can be used to calibrate future carbon emissions.

[0045] Furthermore, the production capacity data includes natural gas production, crude oil production, water production and water injection;

[0046] The energy consumption data include crude oil consumption, diesel consumption, natural gas consumption, flare gas emissions and vent air emissions;

[0047] The types of offshore oil and gas fields include conventional oil fields, conventional gas fields, heavy oil thermal recovery oil fields, low permeability oil fields, low permeability gas fields and high carbon gas fields.

[0048] Furthermore, the preprocessing includes missing value processing, outlier processing, data type conversion and deduplication processing.

[0049] In this embodiment, missing value processing is performed to check whether there are missing values in the data. For example, some hydroelectric power stations may not provide all the information or some information is not recorded. For missing values, you can choose to delete the corresponding records, fill in the missing values (for example, use the mean or median to fill), or use interpolation methods to fill. Outlier processing: Identify and process outliers, which may have a negative impact on the performance of the model. Outliers may be data entry errors or represent real but rare situations. A common processing method is to detect outliers through box plots or Z-score methods, and adjust or delete them according to actual conditions, such as processing points with large deviations in the data. Data type conversion: Ensure the consistency of data types, such as converting text data into numerical data to facilitate subsequent modeling and analysis. Deduplication: Check whether there are duplicate records. If so, the duplicate records need to be deleted to avoid introducing unnecessary deviations in the modeling process.

[0050] Furthermore, the carbon emissions from the multiple production links include carbon emissions from flare gas, carbon emissions from vented air, carbon emissions from crude oil combustion, carbon emissions from diesel combustion and carbon emissions from natural gas combustion.

[0051] In this embodiment, the specific formula for flare gas carbon emissions is as follows:

[0052]

[0053] Among them, E 火炬 is the carbon emission of the flare, V is the gas volume of the flare, is the percentage of carbon dioxide in the flare gas, C 烃 is the hydrocarbon carbon content of the flare gas, η is the carbon oxidation rate of the flare system, is the percentage of methane in the flare gas, GWP is the global warming potential of methane, and C1, C2, C3, IC4, NC4, IC5, NC5, and C6+ are the volume concentrations of carbon-containing compounds in the flare under normal operating conditions in the production of oil and gas fields.

[0054] The specific formula for air carbon emissions is as follows:

[0055] E 工艺放空 =E 处理放空 +E 脱碳直排 +E 火炬冷放空

[0056]

[0057] Among them, N 井口平台 is the number of wellhead platforms, F 井口平台 is the process venting factor of the wellhead platform, N 中心平台 / FPSO is the number of central platforms or FPSOs, F 中心平台 / FPSO is the process venting factor of the central platform or FPSO, GWP is the global warming potential of methane, V CO2 is the volume of carbon dioxide directly discharged from decarbonization, is the volume concentration of carbon dioxide, is the volume of decarbonized direct methane, is the volume concentration of methane, GWP is the global warming potential of methane, V 冷放空 is the volume of cold vent discharge, is the percentage of carbon dioxide in the flare gas, is the percentage of methane in the flare gas.

[0058] The specific formula for carbon emissions from crude oil combustion is as follows:

[0059]

[0060] Among them, M 原油 NCV is the crude oil fuel combustion consumption 原油 is the low calorific value of crude oil, C 原油 is the carbon content per unit calorific value of crude oil, η 原油 is the carbon oxidation rate of crude oil.

[0061] The specific formula for diesel combustion carbon emissions is as follows:

[0062]

[0063] Among them, M 柴油 NCV is the diesel fuel consumption 柴油 is the low calorific value of diesel, C 柴油 is the carbon content per unit calorific value of diesel, η 柴油 is the carbon oxidation rate of diesel.

[0064] The specific formula for carbon emissions from natural gas combustion is as follows:

[0065]

[0066] Among them, V 天然气 is the natural gas fuel combustion consumption, C 天然气 is the carbon content of natural gas, η 天然气 is the carbon oxidation rate of natural gas, and C1, C2, C3, IC4, NC4, IC5, NC5, and C6+ are the volume concentrations of each component in natural gas.

[0067] The calculation formula for carbon emissions in multiple production links is as follows:

[0068] E 总 =E 原油 +E 天然气 +E 火炬 +E 工艺放空 +E 柴油 .

[0069] Furthermore, the BP neural network model is iteratively optimized by a gradient descent algorithm, wherein the hidden layer activation function adopts a tangent S-type transfer function, and the output layer activation function adopts a linear transfer function.

[0070] In this embodiment, the model training adopts a BP neural network. The BP neural network model includes an input layer, a hidden layer, and an output layer. The feature quantity and carbon emissions sorted out by correlation are used as input nodes, and the carbon emissions of offshore oil and gas fields are used as nodes of the output layer. The number of nodes in the hidden layer is preliminarily determined based on the number of nodes in the input layer and the output layer.

[0071] The hidden layer activation function adopts a tangent S-type transfer function, the output layer activation function adopts a linear transfer function, the training function of the BP neural network model adopts a gradient descent BP algorithm training function, and the learning function of the BP neural network model adopts a momentum gradient descent weight and threshold learning function.

[0072] Furthermore, the carbon satellite remote sensing data obtained includes the carbon dioxide concentration in the area where the oil and gas fields are located;

[0073] Calibrate carbon emission predictions based on regional CO2 concentrations:

[0074]

[0075] Where E is the regional carbon emissions, ΔC is the CO2 concentration difference inside and outside the region, h is the height of the atmospheric mixing layer, ρ is the air density, A is the regional area, and Δt is the time interval.

[0076] In this embodiment, the carbon emission correction is obtained by accessing carbon satellite data. The atmospheric carbon dioxide partial pressure in the area where the oil and gas field is located is obtained by accessing carbon satellite data, and the regional carbon dioxide concentration is obtained by the following formula:

[0077]

[0078] in, is the CO2 concentration, pCO2 is the partial pressure of CO2, P 总 is the total gas pressure.

[0079] The total annual carbon emissions are calculated based on the predicted value of calibrated carbon emissions using the following formula:

[0080]

[0081] Where E is the regional carbon emissions and Δt is the time interval.

[0082] Furthermore, the multi-scale fusion algorithm includes: hierarchically fusing the global emission constraints provided by the carbon satellite data with the local prediction values of the BP neural network model.

[0083] Furthermore, the Pearson correlation coefficient analysis method was used to screen the characteristic quantities related to carbon emissions, and the characteristic quantities with the absolute value of the correlation coefficient greater than 0.7 were screened.

[0084] In this embodiment, the correlation analysis adopts the Pearson correlation analysis method, which is defined as the quotient of the product of the covariance and standard deviation between two variables (or the normalized covariance). The specific calculation formula is as follows:

[0085]

[0086] By estimating the covariance and standard deviation of the sample, we can get the Pearson correlation coefficient (of the sample), which is usually represented by the lowercase letter r. The expression of r is as follows:

[0087]

[0088] in, and They represent the sample means of the two respectively.

[0089] When the correlation coefficient is 0, there is no relationship between the two variables X and Y.

[0090] When the value of X increases (decreases), the value of Y also increases (decreases), and the two variables are positively correlated, with a correlation coefficient between 0.00 and 1.00.

[0091] When the value of X increases (decreases) and the value of Y decreases (increases), the two variables are negatively correlated, and the correlation coefficient is between -1.00 and 0.00.

[0092] Then perform feature selection. The following points should be followed when selecting features:

[0093] 1) The larger the absolute value of the correlation coefficient, the stronger the correlation. Generally, a value above 0.7 indicates a strong correlation, and a value below 0.3 indicates a weak correlation.

[0094] 2) If multiple features are highly correlated, only one can be retained to avoid redundancy;

[0095] 3) Choose features that are relevant to your business goals; even if they are not statistically relevant, they may still be meaningful.

[0096] Furthermore, a BP neural network model was constructed, the model was trained using the training set, and the model performance was verified using the test set. The ratio of the training set to the test set was 4:1 or 9:1, and the model performance evaluation indicators included R2, MAE, and RMSE.

[0097] In this example, a machine learning model was built in VS Code using Python. A BP neural network was selected for learning. Cross-validation was used to assess model stability, and hyperparameters were adjusted using grid search or Bayesian optimization. Performance evaluation metrics included R² explained variance, mean absolute error (MAE), and root mean square error (RMSE).

[0098] In summary, the present invention significantly improves the integrity and quality of the data by systematically collecting multidimensional data covering production capacity, energy consumption and oil and gas field types, and adopting interpolation and outlier processing technology. Compared with traditional methods that rely only on a single data source or ignore data cleaning, the present invention can more accurately reflect the actual production situation and provide a reliable basis for carbon emission accounting. It also refines the calculation of carbon emissions for different production links (such as mining, water injection, and flare combustion), avoiding errors caused by simplified assumptions in traditional methods. It also uses machine learning methods to improve the accuracy and timeliness of carbon emission predictions. In the above method, the high resolution and near real-time nature of carbon satellite data can also be used to dynamically reflect the emission changes of oil and gas fields. By combining satellite data with predicted carbon emission data, the prediction accuracy of the model is improved.

[0099] In the aforementioned invention, the existing technology primarily relies on oil and gas field production reports and partial data from ground monitoring points to calculate carbon emissions using a simple emission factor method or mass balance method. However, this approach has significant limitations: the carbon emissions calculated only cover a portion of the production process (such as fuel combustion and flaring) and cannot fully reflect the overall carbon emissions of offshore oil and gas fields. Furthermore, when used in complex production environments (such as heavy oil thermal recovery and high-carbon gas fields), the accuracy of traditional monitoring methods is far inferior to the multi-source data fusion technology used in the present invention.

[0100] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0101] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0102] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0103] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or terminal device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements that are inherent to such process, method, article, or terminal device. In the absence of further restrictions, an element defined by the phrase "comprises a ..." does not exclude the presence of additional identical elements in the process, method, article, or terminal device that includes the element.

[0104] The above is a detailed introduction to the carbon emission prediction method for offshore oil and gas fields provided by this application. Specific examples are used in this article to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method of this application and its core idea; at the same time, for general technical personnel in this field, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on this application.

Claims

1. A method for predicting carbon emissions from offshore oil and gas fields, characterized in that: include: S1. Obtaining raw data of offshore oil and gas fields, including production capacity data, energy consumption data, and oil and gas field types; S2. Preprocessing the acquired raw data to obtain preprocessed data; S3. Calculate the carbon emissions of multiple production links based on the pre-processed data; S4. Based on the calculation results of carbon emissions, the Pearson correlation analysis method is used to screen the characteristic quantities; S5. Inputting the selected feature values into a BP neural network model, wherein the BP neural network model outputs a predicted value of carbon emissions; S6. Obtain carbon satellite remote sensing data, and calibrate the predicted value of carbon emissions of the BP neural network model using a multi-scale fusion algorithm to obtain a calibrated predicted value of carbon emissions.

2. The method for predicting carbon emissions from offshore oil and gas fields according to claim 1, wherein: The production capacity data includes natural gas production, crude oil production, water production and water injection; The energy consumption data include crude oil consumption, diesel consumption, natural gas consumption, flare gas emissions and vent air emissions; The types of offshore oil and gas fields include conventional oil fields, conventional gas fields, heavy oil thermal recovery oil fields, low permeability oil fields, low permeability gas fields and high carbon gas fields.

3. The method for predicting carbon emissions from offshore oil and gas fields according to claim 1, wherein: The preprocessing includes missing value processing, outlier processing, data type conversion and deduplication processing.

4. The method for predicting carbon emissions from offshore oil and gas fields according to claim 1, wherein: The carbon emissions from the multiple production links include carbon emissions from flare gas, carbon emissions from vented air, carbon emissions from crude oil combustion, carbon emissions from diesel combustion and carbon emissions from natural gas combustion.

5. The method for predicting carbon emissions from offshore oil and gas fields according to claim 1, wherein: The BP neural network model is iteratively optimized by a gradient descent algorithm, wherein the hidden layer activation function adopts a tangent S-type transfer function, and the output layer activation function adopts a linear transfer function.

6. The method for predicting carbon emissions from offshore oil and gas fields according to claim 2, wherein: The carbon satellite remote sensing data obtained includes the carbon dioxide concentration in the area where the oil and gas fields are located; Calibrate carbon emission predictions based on regional CO2 concentrations: Where E is the regional carbon emissions, ΔC is the CO2 concentration difference inside and outside the region, h is the height of the atmospheric mixing layer, ρ is the air density, A is the regional area, and Δt is the time interval.

7. The method for predicting carbon emissions from offshore oil and gas fields according to claim 1, wherein: The multi-scale fusion algorithm includes: hierarchically fusing the global emission constraints provided by carbon satellite data with the local prediction values of the BP neural network model.

8. The method for predicting carbon emissions from offshore oil and gas fields according to claim 1, wherein: The Pearson correlation coefficient analysis method was used to screen the characteristic quantities related to carbon emissions, and the characteristic quantities with an absolute value of the correlation coefficient greater than 0.7 were screened.

9. The method for predicting carbon emissions from offshore oil and gas fields according to claim 1, wherein: A BP neural network model was constructed, and the model was trained using the training set. The model performance was verified using the test set. The ratio of the training set to the test set was 4:1 or 9:

1. The model performance evaluation indicators included R2, MAE, and RMSE.

Images