Hybrid model-driven oil and gas station and reservoir combined regulation and operation optimization system

Abstract

The invention belongs to the technical field of oil and gas station libraries, and particularly relates to a hybrid model-driven oil and gas station library joint regulation and operation optimization system, which comprises a construction and model debugging module, a control module, a control module and an optimization module, and is characterized in that the construction and model debugging module is used for firstly carrying out mechanism model establishment and then carrying out model debugging; the optimized operation AI model module is used for firstly establishing a virtual measurement AI model and then operating an optimized AI model; and the mechanism + AI dual-mode driving joint optimization operation module is used for firstly carrying out dual-mode control development and then carrying out dual-model fusion. According to the method, by integrating equipment design parameters, real-time operation data and technological process characteristics, a refined mechanism model containing multiphase flow dynamics and mass and heat transfer laws is constructed, and the problem that a traditional model is insufficient in description of complex working conditions such as emulsion separation and equipment aging is solved.

Description

Technical Field

[0001] This invention relates to the field of oil and gas station technology, specifically to a hybrid model-driven joint control and operation optimization system for oil and gas stations. Background Technology

[0002] With the accelerated global energy transition, the oil and gas industry is placing higher demands on the efficient, energy-saving, and intelligent operation of oil and gas depots. As key hubs for oil and gas gathering, processing, and export, the operational efficiency and stability of oil and gas depots directly impact the security of energy supply and the economic benefits of enterprises. However, traditional oil and gas depot operation and management mainly rely on single mechanistic models or empirical rules, which are difficult to adapt to complex and changing operating conditions and increasingly stringent energy efficiency and environmental protection requirements; specifically, they suffer from the following problems:

[0003] 1. Mechanistic model:

[0004] 1.1 Insufficient Accuracy in Simulating Complex Operating Conditions: Existing mechanistic models for key facilities in stations struggle to accurately describe the mass and heat transfer processes between phases and the dynamic characteristics of equipment when facing complex multiphase flow conditions. For example, when a three-phase separator processes an oil-gas mixture containing a large amount of emulsion or impurities, the mechanistic model based on traditional thermodynamics and fluid mechanics theories shows significant deviations between the simulated values ​​and actual measurements of separation efficiency and the content of each phase component. This results in an inability to accurately guide the adjustment of separator operating parameters, potentially leading to increased oil and gas losses or blockages in downstream equipment.

[0005] 1.2 Delayed Model Parameter Updates: During long-term operation, the physical characteristics of equipment in the station change due to factors such as wear and scaling. However, the parameters of the mechanistic model often fail to be updated in a timely manner to reflect the actual condition of the equipment. Taking an export pump as an example, as the impeller wears, the pump's efficiency curve shifts. If the mechanistic model still uses the initial design parameters, the simulated values ​​of key indicators such as flow rate and pressure will gradually deviate from the actual values ​​on site, affecting the accurate assessment and control of the pump's operating status.

[0006] 2. AI Models

[0007] 2.1 Data Quality and Quantity Limitations: The data acquisition systems of some older oil and gas depots are incomplete, exhibiting problems such as missing data, high noise levels, and poor time synchronization, resulting in low-quality data used for training AI models. For example, key parameters of the heating furnace, such as furnace temperature and fuel gas flow rate, may contain missing or outlier values. This prevents the big data-based virtual metering AI model from fully learning the operating patterns of the heating furnace during training, leading to inaccurate virtual metering results and failing to provide a reliable basis for optimizing the heating furnace's energy efficiency.

[0008] 2.2 Insufficient Model Generalization Ability: AI models trained solely on historical data exhibit poor generalization ability when faced with unprecedented extreme operating conditions or new equipment and processes. For example, when the operating conditions of an external pump suddenly change drastically, such as a significant change in the viscosity or density of the transported medium, and the training data lacks data for such conditions, the operation optimization AI model may be unable to accurately predict the optimal operating range of key parameters, leading to a decrease in pump efficiency or even equipment failure.

[0009] 3. Joint regulation and control

[0010] 3.1 Challenges in Model Integration: Differences in modeling principles, data requirements, and time scales between mechanistic models and AI models make their integration difficult. Mechanistic models focus on the physical nature and causal relationships of equipment, with slower computation speeds but stronger interpretability; AI models rely on data-driven approaches, with faster computation speeds but poorer interpretability. Effectively addressing these differences in real-time performance and accuracy to achieve complementary advantages during joint control is a pressing issue. For example, in the temperature control of a heating furnace, mechanistic models can provide the physical mechanisms of temperature changes, but require significant real-time computation; AI models can quickly provide temperature predictions, but lack a deep understanding of the combustion process, making it difficult for the two to work together to achieve precise temperature control.

[0011] 3.2 Difficulty in Balancing Multiple Objectives: The operation of oil and gas stations requires balancing multiple objectives, including optimal energy efficiency, flexible load, and equipment lifespan. Traditional control technologies often struggle to achieve this balance. For example, pursuing optimal energy efficiency for export pumps may lead to excessively high pump operating pressure, increasing equipment wear and shortening its lifespan. Conversely, considering flexible load to adapt to fluctuations in upstream gas flow may sacrifice some energy efficiency. Current control systems lack effective multi-objective optimization algorithms and models, making it impossible to find the optimal balance point for each objective under different operating conditions.

[0012] Based on the above, a hybrid model-driven joint control and operation optimization system for oil and gas stations and depots is invented. Summary of the Invention

[0013] To address the aforementioned technical problems, according to one aspect of the present invention, the present invention provides the following technical solution:

[0014] A hybrid model-driven joint control and operation optimization system for oil and gas depots includes:

[0015] The mechanism model module is used to receive static and dynamic data from the three-phase separator, external pump, and flow meter at the station, perform simulation calculations on the data, and organize it into a dataset.

[0016] The big data model module is used to receive some dynamic data and data from the dataset in the mechanism model module, and to perform virtual calculations on the data;

[0017] The mechanism + AI dual-mode driven module is used to receive data processed by the mechanism model module and the big data model module, and to optimize the data to achieve production optimization and leakage early warning.

[0018] As a preferred embodiment of the hybrid model-driven joint control and operation optimization system for oil and gas stations described in this invention, the mechanism model module includes:

[0019] The mechanism model building module is used to first collect dynamic and static data, and then combine the gathering and transportation process to build mechanism models of key facilities in the station, and integrate the mechanism models of each key facility.

[0020] The model debugging module is used to first collect on-site production data, then debug key equipment in the station individually, compare and analyze the simulated values ​​of the mechanism model with the actual values ​​on site, and debug the parameters that have a significant impact on the model output results.

[0021] As a preferred embodiment of the hybrid model-driven joint control and operation optimization system for oil and gas stations and depots described in this invention, the mechanism model module includes the following steps:

[0022] S1: Collect dynamic and static data;

[0023] S2: First, combine the gathering and transportation process flow, and then establish the mechanism model of key facilities in the station based on the basic principles; for three-phase separators, establish a mathematical model between separation efficiency and operating parameters based on the flow characteristics, phase balance relationship and separation mechanism of multiphase flow; for external pumps, establish a pump operation characteristic model based on the pump head-flow rate curve and power-flow rate curve; for flow meters, establish the relationship between parameters according to their measurement principle.

[0024] S3: Using a modular modeling approach, the mechanism models of various key facilities are integrated to form an overall process mechanism model of the station, which can simulate the operating status of the station under different working conditions.

[0025] As a preferred embodiment of the hybrid model-driven joint control and operation optimization system for oil and gas stations and depots described in this invention, the mechanism model module further includes the following steps:

[0026] P1: Collect on-site production data, including key indicator data under normal and abnormal operating conditions;

[0027] P2: Conduct individual commissioning of key equipment in the station and compare and analyze the simulated values ​​of the mechanism model with the actual values ​​on site; for the three-phase separator, adjust the key parameters in the model to make the simulated values ​​close to the measured values ​​on site; for the external pump, correct the pump performance curve parameters based on the actual operating data of the pump; for the heating furnace, optimize the parameters of the combustion model.

[0028] P3: Sensitivity analysis is used to identify parameters that have a significant impact on the model output results, and these parameters are adjusted to improve the efficiency of model debugging. At the same time, model error evaluation indicators are established to quantitatively evaluate the accuracy of the model and ensure that the simulation error of the model under different working conditions is controlled within a reasonable range.

[0029] As a preferred embodiment of the hybrid model-driven joint control and operation optimization system for oil and gas stations and depots described in this invention, the big data model module includes:

[0030] The virtual metering AI model building module is used to first develop virtual metering AI algorithms and parameter soft measurement technologies for key parameters of stations and warehouses. Based on actual production data, combined with mechanism simulation to supplement large datasets, it conducts research on AI algorithms based on big data to build a system AI model.

[0031] The operation optimization AI model module is used to first analyze the factors affecting the energy efficiency of external pumps and heating furnaces based on machine learning and big data analysis, then establish an AI model for energy efficiency evaluation and optimization based on decision analysis algorithms, train the operation optimization AI model, and predict the optimal operating range of key parameters under different operating conditions.

[0032] As a preferred embodiment of the hybrid model-driven joint control and operation optimization system for oil and gas stations described in this invention, the big data model module includes the following steps:

[0033] Q1: Develop AI algorithms for virtual metering of key parameters in research and development stations and databases and soft measurement technology for parameters, and analyze the relationship between key parameters and measurable variables for key equipment;

[0034] Q2: Based on actual production data, supplement the large dataset with mechanism simulation; for working conditions with missing data, use mechanism models to simulate and generate virtual data to expand the scale and coverage of training data; and use preprocessing techniques to improve data quality.

[0035] Q3: First, conduct research on AI algorithms based on big data, establish a system AI model, and then, through training the model, realize the virtual measurement of key system parameters;

[0036] Q4: Establish an online update mechanism for the model to dynamically optimize the model based on new data collected in real time, thereby improving the model's adaptability and accuracy.

[0037] As a preferred embodiment of the hybrid model-driven joint control and operation optimization system for oil and gas stations and depots described in this invention, the data model module further includes the following steps:

[0038] M1: Based on machine learning and big data analysis, factors affecting the energy efficiency of external pumps and heating furnaces;

[0039] M2: Employing a combined weighting method, integrating subjective and objective information to determine the weights of each influencing factor, thereby improving the accuracy and reliability of weight calculation; simultaneously, based on decision analysis algorithms, establishing an AI model for energy efficiency evaluation and optimization.

[0040] M3: Using historical operating data and simulation data from the mechanism model, train the operation optimization AI model so that it can predict the optimal operating range of key parameters under different working conditions.

[0041] As a preferred embodiment of the hybrid model-driven joint control and operation optimization system for oil and gas stations described in this invention, the mechanism + AI dual-mode driven module includes:

[0042] The dual-mode control development module is used to first develop dual-mode control technology for joint regulation and equipment operating parameter optimization, then incorporate energy efficiency indicators, flexible load indicators, and equipment reliability indicators into the optimization objectives, and achieve multi-objective collaborative optimization through dual-mode control technology, followed by dynamic adjustment of station and reservoir operating parameters;

[0043] The dual-model fusion module is used to first carry out feature engineering construction, extract key feature variables from the mechanism model and AI model, then perform model optimization, and finally establish a model that runs in conjunction with the dual models.

[0044] As a preferred embodiment of the hybrid model-driven joint control and operation optimization system for oil and gas stations described in this invention, the mechanism + AI dual-mode driving module includes the following steps:

[0045] N1: Develop a dual-mode control technology that combines joint regulation and equipment operating parameter optimization; in which, the mechanistic model provides physical constraints and prior knowledge for the AI ​​model, ensuring that the prediction results of the AI ​​model meet the physical characteristics and process requirements of the equipment; the AI ​​model provides a data-driven supplement to the mechanistic model, improving the prediction accuracy and real-time performance of the model under complex working conditions;

[0046] N2: Establish a multi-objective optimization function, incorporating energy efficiency indicators, flexible load indicators, and equipment reliability indicators into the optimization objectives, and achieve coordinated optimization of multiple objectives through dual-mode control technology;

[0047] N3: Design a dual-mode control strategy. Under normal operating conditions, the AI ​​model is used for real-time regulation, leveraging its advantages of rapid response and data-driven operation to achieve dynamic adjustment of station and warehouse operating parameters. Under abnormal operating conditions or critical decision-making moments, a mechanism model is introduced for verification and correction to ensure the safety and rationality of the regulation strategy.

[0048] As a preferred embodiment of the hybrid model-driven joint control and operation optimization system for oil and gas stations described in this invention, the mechanism + AI dual-mode driving module further includes the following steps:

[0049] E1: To address the discrepancy between the real-time performance and accuracy of key facility mechanisms and AI models, feature engineering will be implemented to extract key feature variables from the mechanism model and the AI ​​model.

[0050] E2: Model optimization is carried out. First, the parameters of the mechanism model and the structure of the AI ​​model are adjusted to make them more consistent in the feature space. Then, on-site comparison and verification are carried out. In actual station operation, the output results of the mechanism model and the AI ​​model are compared in real time to analyze the differences and reasons. The model is optimized based on the on-site feedback data.

[0051] E3: Establish a dual-mode joint operation model and adopt a model fusion algorithm to fuse the prediction results of the mechanism model and the AI ​​model to obtain more accurate and reliable system operation status estimation and control instructions.

[0052] Compared with existing technologies:

[0053] 1. Breakthrough advantages in addressing mechanistic modeling problems

[0054] 1.1 Multi-dimensional data fusion modeling: By integrating equipment design parameters, real-time operating data and process characteristics, a refined mechanism model containing multiphase flow dynamics and mass and heat transfer laws is constructed, which solves the problem that traditional models are insufficient in describing complex working conditions such as emulsion separation and equipment aging.

[0055] 1.2 Dynamic Parameter Adaptive Adjustment: Sensitivity analysis and error quantification assessment are introduced to address parameter drift caused by equipment wear, scaling, etc. Key parameters of the model are dynamically corrected through on-site measured data, reducing the error between simulated and actual values ​​and significantly improving the model's adaptability to changes in equipment status.

[0056] 1.3 Not only is the debugging based on normal operating condition data, but also the measured data of abnormal operating conditions are incorporated. Through modular modeling and system integration, the dynamic characteristics of equipment in multiple scenarios can be accurately simulated, solving the prediction deviation problem of traditional models under extreme operating conditions.

[0057] 2. Core capabilities to overcome AI model bottlenecks

[0058] 2.1 Virtual Data Augmentation Technology: To address the issue of missing data in old stations and warehouses, the missing operating condition data is generated by simulating using a mechanism model. This data is then integrated with actual production data to construct a large-scale training set. Combined with data cleaning and normalization preprocessing, the accuracy of the virtual metering AI model is greatly improved, breaking through the limitations of simply relying on historical data.

[0059] 2.2 Online Dynamic Optimization Algorithm: A real-time model update mechanism is established. Through incremental learning technology, the AI ​​model can automatically adapt to new equipment characteristics or extreme working conditions. The generalization ability is greatly improved compared with the traditional static model, effectively solving the problems of "data cold start" and failure of prediction due to sudden changes in working conditions.

[0060] 2.3 In the operation optimization AI model, the mechanism analysis and data mining are integrated by combining weighting method to accurately quantify the weight of energy efficiency influencing factors, avoid the "black box" defect of the purely data-driven model, so that the optimization strategy is both in line with physical laws and supported by data, and ensures that reliable optimal operating range can still be output under extreme working conditions.

[0061] 3. Innovative breakthroughs in joint regulation technology

[0062] 3.1 Dynamic division of labor mechanism: Under normal working conditions, the AI ​​model takes the lead in real-time control, taking advantage of its data-driven rapid decision-making. Under abnormal working conditions or critical operations, the mechanism model is triggered to verify physical constraints, which solves the contradiction between real-time performance and safety in traditional single models.

[0063] 3.2 Multi-objective optimization function: A three-dimensional objective function including energy efficiency, flexible load and equipment life is constructed, which breaks through the limitation of "secondary problems caused by single-objective optimization" in traditional control technology.

[0064] Unified feature space construction: Extract the physical features of the mechanism model and the data features of the AI ​​model, and form a fused feature vector through principal component analysis to solve the "language barrier" caused by the difference in modeling principles between the two types of models.

[0065] 3.3 Bayesian Fusion Decision: A confidence evaluation system for the mechanistic model and the AI ​​model is established using a Bayesian network. The outputs of the two are dynamically and dynamically fused in real time, which greatly improves the accuracy of joint control compared with a single model and significantly enhances the robustness of the system under complex working conditions.

[0066] 4. Systematic guarantees for technology implementation

[0067] 4.1 Full-process verification system: From single-device debugging of mechanism model to full-system joint debugging of dual-mode joint control, a three-level verification mechanism of "simulation verification → semi-physical test → field pilot" is established to ensure seamless integration of technical solutions with actual production environment.

[0068] 4.2 Intelligent Operation and Maintenance Support: Through real-time monitoring of model errors and AI model performance evaluation, it provides on-site engineers with visual diagnostic tools, solving the pain point of traditional model "optimization relying on human experience" and reducing the cost of technology implementation and maintenance. Attached Figure Description

[0069] Figure 1 This is a schematic diagram of the overall framework of the present invention. Detailed Implementation

[0070] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0071] This invention provides a hybrid model-driven joint control and operation optimization system for oil and gas stations and depots. Please refer to [link / reference]. Figure 1 ,include:

[0072] The mechanism model module is used to receive static and dynamic data from the three-phase separator, external pump, and flow meter at the station, perform simulation calculations on the data, and organize it into a dataset.

[0073] The big data model module is used to receive some dynamic data and data from the dataset in the mechanism model module, and to perform virtual calculations on the data;

[0074] The mechanism + AI dual-mode driven module is used to receive data processed by the mechanism model module and the big data model module, and to optimize the data to achieve production optimization and leakage early warning.

[0075] The mechanism model module:

[0076] The mechanism model building module is used to first collect dynamic and static data, and then combine the gathering and transportation process to build mechanism models of key facilities in the station, and integrate the mechanism models of each key facility.

[0077] The model debugging module is used to first collect on-site production data, then debug key equipment in the station individually, compare and analyze the simulated values ​​of the mechanism model with the actual values ​​on site, and debug the parameters that have a significant impact on the model output results, thereby improving the efficiency of model debugging.

[0078] The mechanism model module includes the following steps:

[0079] S1: Collect dynamic and static data from the station's three-phase separator, external pump, flow meter, etc., including equipment design parameters, material properties, pipeline layout and valve characteristics in the process flow, as well as real-time data such as temperature, pressure, flow rate, and component content at each measuring point;

[0080] S2: First, combine the gathering and transportation process flow, and then, based on the basic principles of thermodynamics, fluid mechanics, mass transfer, and heat transfer, establish a mechanism model for key facilities in the station; for three-phase separators, based on the flow characteristics of multiphase flow, phase equilibrium relationship, and separation mechanism, establish a mathematical model between separation efficiency and operating parameters (such as pressure, temperature, and residence time); for external pumps, based on the pump's head-flow rate curve and power-flow rate curve, establish a pump operating characteristic model, considering the influence of impeller speed, medium viscosity, and other factors on pump performance; for flow meters, based on their measurement principles (such as orifice plate flow meters, turbine flow meters, etc.), establish the relationship between flow rate and parameters such as pressure difference and speed.

[0081] S3: Using a modular modeling approach, the mechanism models of various key facilities are integrated to form an overall process mechanism model of the station, which can simulate the operating status of the station under different working conditions.

[0082] The mechanism model module also includes the following steps:

[0083] P1: Collect on-site production data, including key indicator data under normal and abnormal operating conditions, such as thermodynamic parameters (temperature, pressure, enthalpy), flow rate, pressure, etc.

[0084] P2: Conduct individual commissioning of key equipment at the station, and compare and analyze the simulated values ​​of the mechanism model with the actual values ​​on site; for the three-phase separator, focus on commissioning indicators such as separation efficiency and the content of each phase outlet component, and adjust key parameters in the model (such as phase equilibrium constant, flow resistance coefficient, etc.) to make the simulated values ​​close to the measured values ​​on site; for the external pump, commissioning indicators such as flow rate, head, and power, and correcting the pump performance curve parameters based on the actual operating data of the pump; for the heating furnace, commissioning indicators such as furnace temperature, thermal efficiency, and fuel gas consumption, and optimizing the parameters of the combustion model;

[0085] P3: Sensitivity analysis is used to identify parameters that have a significant impact on the model output results, and these parameters are adjusted to improve the efficiency of model debugging. At the same time, model error evaluation indicators, such as root mean square error and mean absolute error, are established to quantitatively evaluate the accuracy of the model and ensure that the simulation error of the model under different working conditions is controlled within a reasonable range.

[0086] ∑α k =1;

[0087]

[0088] The big data model module includes:

[0089] The virtual metering AI model building module is used to first develop virtual metering AI algorithms and parameter soft measurement technologies for key parameters of stations and warehouses. Based on actual production data, combined with mechanism simulation to supplement large datasets, it conducts research on AI algorithms based on big data to build a system AI model.

[0090] The operation optimization AI model module is used to first analyze the factors affecting the energy efficiency of external pumps and heating furnaces based on machine learning and big data analysis, then establish an AI model for energy efficiency evaluation and optimization based on decision analysis algorithms, train the operation optimization AI model, and predict the optimal operating range of key parameters under different operating conditions.

[0091] The big data model module includes the following steps:

[0092] Q1: Research and development of virtual metering AI algorithms and parameter soft measurement technology for key parameters of the research and development station. For key equipment such as heating furnaces and external pumps, analyze the relationship between key parameters (such as furnace temperature and heat load of heating furnaces, efficiency and shaft power of external pumps, etc.) and measurable variables (such as fuel gas flow rate, inlet and outlet pressure, temperature, current, voltage, etc.).

[0093] Q2: Based on actual production data, supplement the large dataset with mechanism simulation; for working conditions with missing data, use mechanism models to simulate and generate virtual data to expand the scale and coverage of training data; and use data cleaning, normalization and other preprocessing techniques to improve data quality.

[0094] Q3: First, conduct research on AI algorithms based on big data, such as deep learning (neural networks, convolutional neural networks, recurrent neural networks, etc.) and machine learning (support vector machines, random forests, etc.), and establish a system AI model. Then, through training the model, realize the virtual measurement of key system parameters. For example, use the AI ​​model to predict the furnace temperature and heat load in real time based on measurable data such as fuel gas flow rate, air flow rate, and inlet and outlet medium temperature of the heating furnace, and solve the problems of inconvenient installation, measurement lag, or data loss of traditional measuring instruments.

[0095] Q4: Establish an online update mechanism for the model to dynamically optimize the model based on new data collected in real time, thereby improving the model's adaptability and accuracy.

[0096] The data model module also includes the following steps:

[0097] M1: Based on machine learning and big data analysis, this study identifies factors influencing the energy efficiency of external pumps and heating furnaces. For example, factors affecting the energy efficiency of external pumps include impeller speed, viscosity and density of the transported medium, pump inlet and outlet pressure difference, and motor efficiency. Factors affecting the energy efficiency of heating furnaces include fuel gas composition and calorific value, excess air coefficient, furnace temperature, and scaling degree on the heating surfaces. Correlation analysis and principal component analysis are used to determine the influence weight of each factor.

[0098] M2: Employs a combined weighting method (such as combining the analytic hierarchy process (AHP) with the entropy weighting method) to integrate subjective and objective information, determine the weights of each influencing factor, and improve the accuracy and reliability of weight calculation; at the same time, based on decision analysis algorithms (such as multi-objective optimization algorithms, reinforcement learning algorithms, etc.), establishes an AI model for energy efficiency evaluation and optimization.

[0099] M3: By utilizing historical operating data and simulation data from mechanistic models, an AI model for operational optimization is trained, enabling it to predict the optimal operating range of key parameters under different operating conditions. For example, for external pumps, the impeller speed and inlet / outlet valve openings that maximize pump efficiency are predicted based on different flow rates and medium characteristics. For heating furnaces, the optimal fuel gas flow rate and excess air coefficient are predicted based on the medium throughput and target outlet temperature, thereby achieving AI optimization of key facilities and improving energy utilization efficiency.

[0100]

[0101] The mechanism + AI dual-mode driven module includes:

[0102] The dual-mode control development module is used to first develop dual-mode control technology for joint regulation and equipment operating parameter optimization, then incorporate energy efficiency indicators, flexible load indicators, and equipment reliability indicators into the optimization objectives, and achieve multi-objective collaborative optimization through dual-mode control technology, followed by dynamic adjustment of station and reservoir operating parameters;

[0103] The dual-model fusion module is used to first carry out feature engineering construction, extract key feature variables from the mechanism model and AI model, then perform model optimization, and finally establish a model that runs in conjunction with the dual models.

[0104] The mechanism + AI dual-mode driven module includes the following steps:

[0105] N1: Guided by objectives such as optimal energy efficiency of stations and storage facilities and flexible loads, we develop a dual-mode control technology that combines joint regulation and optimization of equipment operating parameters. Among them, the mechanistic model provides physical constraints and prior knowledge for the AI ​​model, ensuring that the prediction results of the AI ​​model meet the physical characteristics and process requirements of the equipment. The AI ​​model provides data-driven supplements to the mechanistic model, improving the prediction accuracy and real-time performance of the model under complex operating conditions.

[0106] N2: Establish a multi-objective optimization function, incorporating energy efficiency indicators (such as energy consumption per unit of oil and gas processing, fuel gas utilization rate, etc.), flexible load indicators (such as adaptability to upstream gas fluctuations, adjustment range of external flow, etc.), and equipment reliability indicators (such as equipment operating life, failure rate, etc.) into the optimization objectives, and achieve collaborative optimization of multiple objectives through dual-mode control technology;

[0107] N3: Design a dual-mode control strategy. Under normal operating conditions, the AI ​​model is used for real-time regulation, leveraging its advantages of rapid response and data-driven operation to achieve dynamic adjustment of station and warehouse operating parameters. Under abnormal operating conditions or critical decision-making moments, a mechanism model is introduced for verification and correction to ensure the safety and rationality of the regulation strategy.

[0108] The mechanism + AI dual-mode driven module also includes the following steps:

[0109] E1: To address the discrepancy in real-time performance and accuracy between key facility mechanisms and AI models, feature engineering is implemented to extract key feature variables from both the mechanism model and the AI ​​model, such as thermodynamic parameters and equipment structural parameters in the mechanism model, and data features and prediction error features in the AI ​​model, to establish a unified feature space.

[0110] E2: Model optimization is carried out. First, the parameters of the mechanism model and the structure of the AI ​​model are adjusted to make them more consistent in the feature space. Then, on-site comparison and verification are carried out. In actual station operation, the output results of the mechanism model and the AI ​​model are compared in real time to analyze the differences and reasons. The model is optimized based on the on-site feedback data.

[0111] E3: Establish a dual-mode joint operation model and adopt model fusion algorithms (such as weighted average fusion, Bayesian fusion, etc.) to fuse the prediction results of the mechanism model and the AI ​​model to obtain more accurate and reliable system operation status estimates and control commands. Taking the heating furnace and external pump system as examples, by fusing the understanding of the combustion process and pump operation physical mechanism of the mechanism model and the rapid analysis of real-time data by the AI ​​model, the collaborative optimization and joint control of the "mechanism + AI" model can be realized, thereby improving the energy saving and consumption reduction effect and operation stability of the station and warehouse.

[0112]

[0113] In practical use, the specific steps are as follows:

[0114] S1: First, collect dynamic and static data through the mechanism model building module, then build mechanism models of key facilities in the station in combination with the gathering and transportation process, and then integrate the mechanism models of each key facility.

[0115] S11: Collect dynamic and static data;

[0116] S12: First, combine the gathering and transportation process flow, and then establish the mechanism model of key facilities in the station based on the basic principles; for three-phase separators, establish a mathematical model between separation efficiency and operating parameters based on the flow characteristics, phase balance relationship and separation mechanism of multiphase flow; for external pumps, establish a pump operation characteristic model based on the pump head-flow rate curve and power-flow rate curve; for flow meters, establish the relationship between parameters according to their measurement principle.

[0117] S13: Using a modular modeling approach, the mechanism models of various key facilities are integrated to form an overall process mechanism model of the station, which can simulate the operating status of the station under different working conditions.

[0118] S2: First, collect on-site production data through the model debugging module, then debug key equipment in the station separately, compare and analyze the simulated values ​​of the mechanism model with the actual values ​​on site, then determine the parameters that have a significant impact on the model output results, and debug these parameters to improve the efficiency of model debugging.

[0119] S21: Collect on-site production data, including key indicator data under normal and abnormal operating conditions;

[0120] S22: Conduct individual commissioning of key equipment in the station and compare and analyze the simulated values ​​of the mechanism model with the actual values ​​on site; for the three-phase separator, adjust the key parameters in the model to make the simulated values ​​close to the measured values ​​on site; for the external pump, correct the pump performance curve parameters based on the actual operating data of the pump; for the heating furnace, optimize the parameters of the combustion model.

[0121] S23: Use sensitivity analysis to identify parameters that have a significant impact on the model output results, and adjust these parameters to improve the efficiency of model debugging; at the same time, establish model error evaluation indicators to quantitatively evaluate the accuracy of the model and ensure that the simulation error of the model under different working conditions is controlled within a reasonable range.

[0122] S3: Virtual metering AI model building module, used to first develop virtual metering AI algorithms and parameter soft measurement technology for key station and warehouse parameters, then supplement the big data dataset based on actual production data and mechanism simulation, and then carry out AI algorithm research based on big data to build a system AI model;

[0123] S31: Research and development of virtual metering AI algorithms and parameter soft measurement technology for key parameters of research and development stations, for key equipment, to analyze the relationship between key parameters and measurable variables;

[0124] S32: Based on actual production data, supplement the large dataset with mechanism simulation; for working conditions with missing data, use mechanism models to simulate and generate virtual data to expand the scale and coverage of training data; and use preprocessing techniques to improve data quality.

[0125] S33: First, conduct research on AI algorithms based on big data, establish a system AI model, and then, through training the model, realize the virtual measurement of key system parameters;

[0126] S34: Establish an online update mechanism for the model to dynamically optimize the model based on new data collected in real time, thereby improving the model's adaptability and accuracy;

[0127] S4: Run optimization AI model module, which is used to first analyze the factors affecting the energy efficiency of external pumps and heating furnaces based on machine learning and big data, then establish an AI model for energy efficiency evaluation and optimization based on decision analysis algorithm, and then train the run optimization AI model so that it can predict the optimal operating range of key parameters under different working conditions;

[0128] S41: Based on machine learning and big data analysis, analyze the factors affecting the energy efficiency of external pumps and heating furnaces;

[0129] S42: Employing a combined weighting method, integrating subjective and objective information, the weights of each influencing factor are determined, improving the accuracy and reliability of weight calculation; simultaneously, based on decision analysis algorithms, an AI model for energy efficiency evaluation and optimization is established.

[0130] S43: Using historical operating data and simulation data from mechanism models, train an AI model for operation optimization so that it can predict the optimal operating range of key parameters under different operating conditions;

[0131] S5: Dual-mode control development module, used to first develop dual-mode control technology for joint regulation and equipment operating parameter optimization, then incorporate energy efficiency indicators, flexible load indicators, and equipment reliability indicators into the optimization objectives, and achieve multi-objective collaborative optimization through dual-mode control technology, followed by dynamic adjustment of station and depot operating parameters;

[0132] S51: Develop a dual-mode control technology that combines joint regulation and equipment operating parameter optimization; in which, the mechanistic model provides physical constraints and prior knowledge for the AI ​​model, ensuring that the prediction results of the AI ​​model meet the physical characteristics and process requirements of the equipment; the AI ​​model provides a data-driven supplement to the mechanistic model, improving the prediction accuracy and real-time performance of the model under complex working conditions;

[0133] S52: Establish a multi-objective optimization function, incorporate energy efficiency indicators, flexible load indicators, and equipment reliability indicators into the optimization objectives, and achieve coordinated optimization of multiple objectives through dual-mode control technology;

[0134] S53: Design a dual-mode control strategy. Under normal operating conditions, the AI ​​model is used for real-time regulation, leveraging its advantages of rapid response and data-driven operation to achieve dynamic adjustment of station and depot operating parameters. Under abnormal operating conditions or critical decision-making moments, a mechanism model is introduced for verification and correction to ensure the safety and rationality of the regulation strategy.

[0135] S6: Dual-model fusion module, used to first carry out feature engineering construction, extract key feature variables from the mechanism model and AI model, then perform model optimization, and finally establish a model for joint operation of the two models;

[0136] S61: To address the discrepancy between the real-time performance and accuracy of key facility mechanisms and AI models, feature engineering will be implemented to extract key feature variables from the mechanism model and the AI ​​model.

[0137] S62: Perform model optimization. First, adjust the parameters of the mechanism model and the structure of the AI ​​model to make them more consistent in the feature space. Then, conduct on-site comparison and verification. In actual station operation, compare the output results of the mechanism model and the AI ​​model in real time, analyze the differences and reasons between them, and optimize the model based on the on-site feedback data.

[0138] S63: Establish a dual-mode joint operation model, and use a model fusion algorithm to fuse the prediction results of the mechanism model and the AI ​​model to obtain more accurate and reliable system operation status estimation and control instructions.

[0139] Although the present invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the disclosed embodiments can be combined with each other in any manner. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A hybrid model-driven joint control and operation optimization system for oil and gas stations and depots, characterized in that, include: The mechanism model module is used to receive static and dynamic data from the three-phase separator, external pump, and flow meter at the station, perform simulation calculations on the data, and organize it into a dataset. The big data model module is used to receive some dynamic data and data from the dataset in the mechanism model module, and to perform virtual calculations on the data; The mechanism + AI dual-mode driven module is used to receive data processed by the mechanism model module and the big data model module, and to optimize the data to achieve production optimization and leakage early warning.

2. The hybrid model-driven joint control and operation optimization system for oil and gas stations and depots according to claim 1, characterized in that, The mechanism model module: The mechanism model building module is used to first collect dynamic and static data, and then combine the gathering and transportation process to build mechanism models of key facilities in the station, and integrate the mechanism models of each key facility. The model debugging module is used to first collect on-site production data, then debug key equipment in the station individually, compare and analyze the simulated values ​​of the mechanism model with the actual values ​​on site, and debug the parameters that have a significant impact on the model output results. The big data model module includes: The virtual metering AI model building module is used to first develop virtual metering AI algorithms and parameter soft measurement technologies for key parameters of stations and warehouses. Based on actual production data, combined with mechanism simulation to supplement large datasets, it conducts research on AI algorithms based on big data to build a system AI model. The operation optimization AI model module is used to first analyze the factors affecting the energy efficiency of external pumps and heating furnaces based on machine learning and big data analysis, then establish an AI model for energy efficiency evaluation and optimization based on decision analysis algorithms, train the operation optimization AI model, and predict the optimal operating range of key parameters under different operating conditions.

3. The hybrid model-driven joint control and operation optimization system for oil and gas stations and depots according to claim 2, characterized in that, The mechanism model module includes the following steps: S1: Collect dynamic and static data; S2: First, combine the gathering and transportation process flow, and then establish the mechanism model of key facilities in the station based on the basic principles; for three-phase separators, establish a mathematical model between separation efficiency and operating parameters based on the flow characteristics, phase balance relationship and separation mechanism of multiphase flow; for external pumps, establish a pump operation characteristic model based on the pump head-flow rate curve and power-flow rate curve; for flow meters, establish the relationship between parameters according to their measurement principle. S3: Using a modular modeling approach, the mechanism models of various key facilities are integrated to form an overall process mechanism model of the station, which can simulate the operating status of the station under different working conditions.

4. The hybrid model-driven joint control and operation optimization system for oil and gas stations and depots according to claim 2, characterized in that, The mechanism model module also includes the following steps: P1: Collect on-site production data, including key indicator data under normal and abnormal operating conditions; P2: Conduct individual commissioning of key equipment in the station and compare and analyze the simulated values ​​of the mechanism model with the actual values ​​on site; for the three-phase separator, adjust the key parameters in the model to make the simulated values ​​close to the measured values ​​on site; for the external pump, correct the pump performance curve parameters based on the actual operating data of the pump; for the heating furnace, optimize the parameters of the combustion model. P3: Sensitivity analysis is used to identify parameters that have a significant impact on the model output results, and these parameters are adjusted to improve the efficiency of model debugging. At the same time, model error evaluation indicators are established to quantitatively evaluate the accuracy of the model and ensure that the simulation error of the model under different working conditions is controlled within a reasonable range.

5. The hybrid model-driven joint control and operation optimization system for oil and gas stations and depots according to claim 4, characterized in that, The big data model module includes the following steps: Q1: Develop AI algorithms for virtual metering of key parameters in research and development stations and databases and soft measurement technology for parameters, and analyze the relationship between key parameters and measurable variables for key equipment; Q2: Based on actual production data, supplement the large dataset with mechanism simulation; for working conditions with missing data, use mechanism models to simulate and generate virtual data to expand the scale and coverage of training data; and use preprocessing techniques to improve data quality. Q3: First, conduct research on AI algorithms based on big data, establish a system AI model, and then, through training the model, realize the virtual measurement of key system parameters; Q4: Establish an online update mechanism for the model to dynamically optimize the model based on new data collected in real time, thereby improving the model's adaptability and accuracy.

6. The hybrid model-driven joint control and operation optimization system for oil and gas stations and depots according to claim 1, characterized in that, The data model module also includes the following steps: M1: Based on machine learning and big data analysis, factors affecting the energy efficiency of external pumps and heating furnaces; M2: Employing a combined weighting method, integrating subjective and objective information to determine the weights of each influencing factor, thereby improving the accuracy and reliability of weight calculation; simultaneously, based on decision analysis algorithms, establishing an AI model for energy efficiency evaluation and optimization. M3: Using historical operating data and simulation data from the mechanism model, train the operation optimization AI model so that it can predict the optimal operating range of key parameters under different working conditions.

7. The hybrid model-driven joint control and operation optimization system for oil and gas stations and depots according to claim 1, characterized in that, The mechanism + AI dual-mode driven module includes: The dual-mode control development module is used to first develop dual-mode control technology for joint regulation and equipment operating parameter optimization, then incorporate energy efficiency indicators, flexible load indicators, and equipment reliability indicators into the optimization objectives, and achieve multi-objective collaborative optimization through dual-mode control technology, followed by dynamic adjustment of station and reservoir operating parameters; The dual-model fusion module is used to first carry out feature engineering construction, extract key feature variables from the mechanism model and AI model, then perform model optimization, and finally establish a model that runs in conjunction with the dual models.

8. The hybrid model-driven joint control and operation optimization system for oil and gas stations and depots according to claim 7, characterized in that, The mechanism + AI dual-mode driven module includes the following steps: N1: Develop a dual-mode control technology that combines joint regulation and equipment operating parameter optimization; in which, the mechanistic model provides physical constraints and prior knowledge for the AI ​​model, ensuring that the prediction results of the AI ​​model meet the physical characteristics and process requirements of the equipment; the AI ​​model provides a data-driven supplement to the mechanistic model, improving the prediction accuracy and real-time performance of the model under complex working conditions; N2: Establish a multi-objective optimization function, incorporating energy efficiency indicators, flexible load indicators, and equipment reliability indicators into the optimization objectives, and achieve coordinated optimization of multiple objectives through dual-mode control technology; N3: Design a dual-mode control strategy. Under normal operating conditions, the AI ​​model is used for real-time regulation, leveraging its advantages of rapid response and data-driven operation to achieve dynamic adjustment of station and warehouse operating parameters. Under abnormal operating conditions or critical decision-making moments, a mechanism model is introduced for verification and correction to ensure the safety and rationality of the regulation strategy.

9. The hybrid model-driven joint control and operation optimization system for oil and gas stations and depots according to claim 7, characterized in that, The mechanism + AI dual-mode driven module also includes the following steps: E1: To address the discrepancy between the real-time performance and accuracy of key facility mechanisms and AI models, feature engineering will be implemented to extract key feature variables from the mechanism model and the AI ​​model. E2: Model optimization is carried out. First, the parameters of the mechanism model and the structure of the AI ​​model are adjusted to make them more consistent in the feature space. Then, on-site comparison and verification are carried out. In actual station operation, the output results of the mechanism model and the AI ​​model are compared in real time to analyze the differences and reasons. The model is optimized based on the on-site feedback data. E3: Establish a dual-mode joint operation model and adopt a model fusion algorithm to fuse the prediction results of the mechanism model and the AI ​​model to obtain more accurate and reliable system operation status estimation and control instructions.

Images