Automatic blending control method for oil field fracturing fluid

Abstract

The invention belongs to the field of data processing, and particularly relates to an automatic blending control method for oil field fracturing fluid, which comprises the following steps: acquiring blending production data of the oil field fracturing fluid; performing real-time prediction on the mixed production data through a concentration prediction model to obtain a concentration prediction value of the oil field fracturing fluid; the concentration prediction model is obtained by training actual fracturing fluid concentration values obtained under different production conditions; the actual concentration value of the fracturing fluid comes from detection data of a plurality of pipe network nodes in a production environment; real-time matching information of the oil field fracturing fluid is generated based on the concentration predicted value, and automatic mixing control over the oil field fracturing fluid is executed through the real-time matching information. According to the method, the precision, efficiency and quality of fracturing fluid mixing can be improved, and the adaptability to a complex production environment is enhanced.

Description

Technical Field

[0001] The present application belongs to the field of data processing, and in particular relates to an automatic mixing control method for oilfield fracturing fluid. Background Art

[0002] Oilfield fracturing fluid is a liquid with special properties used in the process of oilfield exploitation, especially in hydraulic fracturing operations. The fracturing fluid is injected into the oil layer under high pressure, and the energy it carries is used to break the oil layer rock, forming cracks with certain conductivity, thereby creating a good channel for subsequent oil and other resources to flow from the formation to the wellbore. The fracturing fluid can carry proppants (such as quartz sand, etc.) into the newly formed cracks. When the pressure is released, the proppants can support the cracks, prevent the cracks from closing, maintain their conductivity, and allow oil and gas to flow continuously and smoothly into the wellbore to be exploited.

[0003] At present, oilfield fracturing fluids are generally simpler liquids such as clean water or salt water. Their main purpose is to be injected into the formation under high pressure before formal fracturing, so as to break the formation rock and create preliminary fractures, thereby creating conditions for the subsequent entry of sand-carrying fluids. In related technologies, if the concentration of the raw materials themselves (such as water, basic polymer solutions, etc.) fluctuates, it may not be possible to accurately perceive and adjust them in a timely manner, resulting in unstable concentration of the final mixed fracturing fluid, affecting its overall quality and the effect of subsequent fracturing operations. In particular, there is a certain time difference from monitoring the data anomaly to the system making corresponding adjustments. In the case of a large mixing flow rate of the fracturing fluid and a complex process, this lag may result in a large number of fracturing fluids that do not meet the requirements being mixed, making it difficult to ensure the stable performance of each batch of fracturing fluids.

[0004] Therefore, in order to solve at least one of the above technical problems, it is urgent to propose a new automatic mixing control solution for oilfield fracturing fluid. Summary of the invention

[0005] The present application provides an automatic mixing control method for oilfield fracturing fluid, which is used to improve the accuracy, efficiency and quality of fracturing fluid mixing and enhance the adaptability to complex production environments.

[0006] In a first aspect, the present application provides an automatic mixing control method for oilfield fracturing fluid, the method comprising:

[0007] Acquire the mixed production data of oilfield fracturing fluid; the mixed production data at least includes: equipment status, environmental factors, production parameters; the production parameters at least include: temperature, conductivity, initial concentration, liquid flow;

[0008] The mixed production data is predicted in real time by a concentration prediction model to obtain a concentration prediction value of the oilfield fracturing fluid; the concentration prediction model is trained using actual concentration values ​​of the fracturing fluid obtained under different production conditions; the actual concentration value of the fracturing fluid comes from detection data of multiple pipe network nodes in the production environment;

[0009] Real-time proportioning information of the oilfield fracturing fluid is generated based on the concentration prediction value, and automatic mixing control of the oilfield fracturing fluid is performed through the real-time proportioning information.

[0010] In a second aspect, an embodiment of the present application provides an automatic mixing control device for oilfield fracturing fluid, the device comprising:

[0011] An acquisition unit is configured to acquire mixed production data of oilfield fracturing fluid; the mixed production data at least includes: equipment status, environmental factors, production parameters; the production parameters at least include: temperature, conductivity, initial concentration, liquid flow rate;

[0012] The prediction unit is configured to perform real-time prediction on the mixed production data through a concentration prediction model to obtain a concentration prediction value of the oilfield fracturing fluid; the concentration prediction model is trained using actual concentration values ​​of the fracturing fluid obtained under different production conditions; the actual concentration value of the fracturing fluid comes from detection data of multiple pipe network nodes in the production environment;

[0013] The execution unit is configured to generate real-time ratio information of the oilfield fracturing fluid based on the concentration prediction value, and execute automatic mixing control of the oilfield fracturing fluid through the real-time ratio information.

[0014] In a third aspect, an embodiment of the present application provides a computing device, the computing device comprising:

[0015] at least one processor, memory, and input-output unit;

[0016] The memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute the automatic mixing control method for oilfield fracturing fluid of the first aspect.

[0017] In a fourth aspect, a computer-readable storage medium is provided, which includes instructions. When the instructions are executed on a computer, the computer executes the automatic mixing control method for oilfield fracturing fluid according to the first aspect.

[0018] In the technical solution provided in the embodiment of the present application, first, the mixed production data of the oilfield fracturing fluid is obtained; the mixed production data includes at least: equipment status, environmental factors, production parameters; the production parameters include at least: temperature, conductivity, initial concentration, liquid flow rate. Then, the mixed production data is predicted in real time through the concentration prediction model to obtain the concentration prediction value of the oilfield fracturing fluid; the concentration prediction model is trained using the actual concentration value of the fracturing fluid obtained under different production conditions; the actual concentration value of the fracturing fluid comes from the detection data of multiple pipeline nodes in the production environment. Finally, the real-time proportioning information of the oilfield fracturing fluid is generated based on the concentration prediction value, and the automatic mixing control of the oilfield fracturing fluid is performed through the real-time proportioning information. In summary, the automatic mixing control method of the oilfield fracturing fluid helps to improve the accuracy, efficiency and quality of the fracturing fluid mixing through many advantages, enhances the adaptability to complex production environments, and promotes the intelligent development of production management, which has positive and important significance for oilfield mining operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0020] Figure 1 It is a flow chart of an automatic mixing control method of an oilfield fracturing fluid according to an embodiment of the present application;

[0021] Figure 2 It is a schematic diagram of the principle of an automatic mixing control method for oilfield fracturing fluid according to an embodiment of the present application;

[0022] Figure 3 It is a structural schematic diagram of an automatic mixing control device for oilfield fracturing fluid in an embodiment of the present application;

[0023] Figure 4 It is a structural schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.

[0026] Oilfield fracturing fluid is a liquid with special properties used in the process of oilfield exploitation, especially in hydraulic fracturing operations. The fracturing fluid is injected into the oil layer under high pressure, and the energy it carries is used to break the oil layer rock, forming cracks with certain conductivity, thereby creating a good channel for subsequent oil and other resources to flow from the formation to the wellbore. The fracturing fluid can carry proppants (such as quartz sand, etc.) into the newly formed cracks. When the pressure is released, the proppants can support the cracks, prevent the cracks from closing, maintain their conductivity, and allow oil and gas to flow continuously and smoothly into the wellbore to be exploited.

[0027] At present, oilfield fracturing fluids are generally simpler liquids such as clean water or salt water. Their main purpose is to be injected into the formation under high pressure before formal fracturing, so as to break the formation rock and create preliminary fractures, and create conditions for the subsequent entry of sand-carrying fluids. Sand-carrying fluid is one of the core parts of the fracturing fluid system. In addition to water-based solutions, it also contains various additives, such as thickeners (natural polymers such as guar gum or synthetic polymers, which can increase the viscosity of the liquid to facilitate better carrying of proppants), drag reducers (reducing the resistance of liquid flow in pipelines and formations), anti-swelling agents (preventing clay minerals from swelling and blocking cracks when they encounter water), and other functional additives to ensure that the proppant can be effectively carried and placed in the appropriate fracture position. Displacement fluid is mainly used to displace the residual sand-carrying fluid in the wellbore into the formation fractures to ensure that the proppant can be fully distributed in the formation fractures. Commonly used are simpler liquids such as clean water, but sometimes a small amount of necessary additives are added according to actual conditions.

[0028] In the related technology, if the concentration of the raw materials themselves (such as water, basic polymer solution, etc.) fluctuates, it may not be possible to accurately sense and make corresponding adjustments in a timely manner, resulting in unstable concentration of the final mixed fracturing fluid, affecting its overall quality and the effect of subsequent fracturing operations. In particular, there is a certain time difference from monitoring data anomalies to the system making corresponding adjustments. In the case of a large flow rate of fracturing fluid mixing and a complex process, this lag may result in a large number of fracturing fluids that do not meet the requirements being mixed, making it difficult to ensure the stable performance of each batch of fracturing fluid.

[0029] Therefore, in order to solve at least one of the above technical problems, it is urgent to propose a new automatic mixing control solution for oilfield fracturing fluid.

[0030] In order to solve at least one of the above technical problems, an embodiment of the present application provides an automatic mixing control method for oilfield fracturing fluid.

[0031] Specifically, first, the mixed production data of the oilfield fracturing fluid is obtained; the mixed production data includes at least: equipment status, environmental factors, production parameters; the production parameters include at least: temperature, conductivity, initial concentration, liquid flow rate. Then, the mixed production data is predicted in real time through the concentration prediction model to obtain the concentration prediction value of the oilfield fracturing fluid; the concentration prediction model is trained using the actual concentration value of the fracturing fluid obtained under different production conditions; the actual concentration value of the fracturing fluid comes from the detection data of multiple pipeline nodes in the production environment. Finally, the real-time proportioning information of the oilfield fracturing fluid is generated based on the concentration prediction value, and the automatic mixing control of the oilfield fracturing fluid is performed through the real-time proportioning information.

[0032] In the technical solution of the present application, firstly, by considering the equipment status, environmental factors and multiple key production parameters (such as temperature, conductivity, initial concentration, liquid flow rate, etc.), various factors that may affect the concentration of fracturing fluid are comprehensively covered. Compared with the traditional method that only focuses on some factors, this method can more accurately grasp the complex situation in actual production, so that the concentration prediction value is closer to the actual situation, which helps to generate more accurate real-time ratio information and improve the mixing accuracy of fracturing fluid. The concentration prediction model in the embodiment of the present application is trained by the actual concentration value of fracturing fluid obtained under different production conditions. These actual concentration values ​​are derived from the detection data of multiple pipe network nodes in the production environment, which means that the model learns the data law under the real production scene. Compared with the model based on theoretical assumptions or limited simulation data training, it can better adapt to the changing working conditions of the oil field site, and then more accurately predict the concentration, and provide a strong guarantee for realizing accurate automatic mixing control. Secondly, it is possible to predict the mixed production data in real time, which makes it possible to obtain the concentration prediction value at any time according to the actual situation at the moment during the mixing process. When faced with fluctuations in raw material concentration, sudden equipment failures, or sudden changes in environmental factors, the system can respond quickly and generate corresponding real-time ratio information in a timely manner to avoid the production of a large amount of fracturing fluid that does not meet the requirements due to feedback lag and other problems, ensuring that the mixing process is always under effective control. Based on the real-time concentration prediction value, real-time ratio information is generated to achieve dynamic regulation of the mixing process. Unlike the preset fixed ratio method, it can flexibly adjust the ratio parameters such as the amount of each raw material added according to the real-time changes in the production process, better adapt to the mixing requirements under different stages and different working conditions, ensure the quality stability of the fracturing fluid, and help improve the success rate and efficiency of the entire fracturing operation. Again, the combination of accurate concentration prediction and real-time automatic mixing control effectively reduces the probability of producing unqualified fracturing fluid due to inaccurate ratios. It reduces the occurrence of rework and re-mixing, avoids resource waste (such as raw material waste, time waste, etc.), improves overall production efficiency, and also ensures that each batch of fracturing fluid put into the fracturing operation can meet the corresponding quality standards, which helps to improve the effect of the fracturing operation and ultimately improve the recovery rate of the oil field. This automated mixing control method reduces the operational errors that may occur during manual intervention, and can more efficiently coordinate the relationship between equipment, raw materials, and various production parameters. The entire production process is smoother and more orderly, and the links between each link are more closely connected, which not only improves the production quality of a single batch of fracturing fluid, but also helps to optimize the overall process flow of oilfield fracturing fluid production in the long run and achieve large-scale, high-quality production operations. Next, environmental factors are included in the consideration of mixed production data, so that the control method can adapt to complex and changeable production environments such as different seasons, different regions, and different climates.For example, the properties of raw materials and the operating status of equipment will be different in high-temperature and low-temperature environments. This method can automatically adjust the ratio according to these environmental differences to ensure that the mixing of fracturing fluid is not disturbed by excessive changes in the external environment, and always maintains stable and reliable performance, expanding its applicability in various oilfield environments. Monitoring the equipment status and incorporating it into the control process as an influencing factor means that when the equipment is aging, has a minor fault, or the operating parameters fluctuate, the system can still adjust the mixing ratio according to the actual status of the equipment, ensuring that qualified fracturing fluid can be produced as much as possible when the equipment is not in the best condition, improving the ability of the entire production system to cope with equipment uncertainties and enhancing the stability and reliability of the production process. Finally, throughout the process, the detection data from multiple pipeline nodes in the production environment are continuously collected and utilized, and a large amount of production data covering different working conditions will be accumulated over time. These data can not only be used for the continuous training and optimization of the current concentration prediction model to continuously improve its performance, but also provide data support for subsequent production process improvements, equipment maintenance plan formulation, quality control, and other aspects, to achieve data-driven refined management and optimization upgrades. With the help of this data-driven automatic mixing control method, enterprises can gradually build an intelligent production decision-making system. By analyzing a large amount of historical data and real-time data, they can explore more valuable production rules and potential problems, provide a scientific basis for further optimizing the production of oilfield fracturing fluids and improving overall production efficiency, and promote oilfield production to develop in a more intelligent and efficient direction.

[0033] In summary, the automatic mixing control method of oilfield fracturing fluid can help improve the accuracy, efficiency and quality of fracturing fluid mixing through many advantages, enhance the adaptability to complex production environments, and promote the intelligent development of production management, which has positive and important significance for oilfield exploitation operations.

[0034] The automatic mixing control scheme of oilfield fracturing fluid provided in the embodiment of the present application can be executed by an electronic device, which can be a server, a server cluster, or a cloud server. The electronic device can also be a terminal device such as a mobile phone, a computer, a tablet computer, a wearable device, or a dedicated device (such as a dedicated terminal device with an automatic mixing control system for oilfield fracturing fluid, etc.). In an optional embodiment, a service program for executing the automatic mixing control scheme of oilfield fracturing fluid can be installed on the electronic device.

[0035] Figure 1 A schematic diagram of an automatic mixing control method for oilfield fracturing fluid provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method comprises the following steps:

[0036] 101, obtaining the mixing production data of the oilfield fracturing fluid;

[0037] 102, using a concentration prediction model, real-time prediction is performed on the mixed production data to obtain a concentration prediction value of the oilfield fracturing fluid;

[0038] 103 , generating real-time proportioning information of the oilfield fracturing fluid based on the concentration prediction value, and performing automatic mixing control of the oilfield fracturing fluid through the real-time proportioning information.

[0039] In the embodiment of the present application, the mixing production data at least includes: equipment status, environmental factors, and production parameters. Further optionally, the production parameters at least include: temperature, conductivity, initial concentration, and liquid flow rate.

[0040] Specifically, as an optional embodiment, 101, obtain the mixing production data of the oilfield fracturing fluid. The equipment status data in the mixing production data, such as the status of the delivery pump. That is, the speed of the delivery pump, which directly affects the delivery speed and flow rate of the raw materials. Different speeds will cause the amount of raw materials entering the mixing system to change, thereby affecting the ratio of the fracturing fluid. In addition, the working pressure of the delivery pump is also a key parameter. Abnormal pressure may mean that there are hidden dangers of failure of the pump body or problems such as pipeline blockage, which will interfere with the normal supply of raw materials and mixing process.

[0041] Agitator parameters: The stirring speed of the agitator determines the uniformity of the raw materials in the mixing container. The appropriate stirring speed can ensure that the components are fully and evenly mixed, while too fast or too slow stirring speed may lead to poor mixing effect. The running time of the agitator is also important. It is related to the entire mixing cycle. Too long or too short running time may affect the quality of the final fracturing fluid.

[0042] The valve status is used to indicate the opening and closing of the valve to control the flow direction of the raw materials and whether each part of the raw materials can enter the mixing process according to the predetermined ratio. The opening of the valve accurately controls the flow of the raw materials. If the valve opening is inaccurate, it will cause deviations in the amount of raw materials added, affecting the concentration ratio of the fracturing fluid.

[0043] Equipment fault code. Modern oilfield fracturing fluid mixing equipment usually has a fault self-diagnosis function, which will generate corresponding fault codes to indicate equipment problems. For example, a sensor failure, motor overload, etc. will have corresponding code records. These fault information reflects the current abnormal state of the equipment and will have a direct or indirect impact on the mixing production. It needs to be considered in the automatic mixing control process in order to adjust the ratio strategy or arrange maintenance in time.

[0044] Equipment maintenance records, equipment maintenance status, such as the last maintenance time, replacement parts information, etc., are also closely related to the current performance status of the equipment. Equipment that has just undergone maintenance may be in a good operating state, while equipment that has not been maintained for a long time may have problems such as performance degradation and parameter instability. Understanding these maintenance records can help more accurately estimate the impact of equipment on mixing production.

[0045] Ambient temperature has a significant impact on the physical and chemical properties of raw materials. For example, in a low-temperature environment, some raw materials may increase in viscosity and deteriorate in fluidity, affecting their transportation in pipelines and mixing effects in mixing containers; while a high-temperature environment may accelerate the chemical reaction rate of certain components in the raw materials, change the content of effective components in the raw materials or cause their stability to decrease, thereby affecting the final performance and concentration of the fracturing fluid.

[0046] Humidity will affect the moisture content of raw materials. For some moisture-sensitive raw materials, high humidity may cause them to absorb moisture, change their own concentration or trigger chemical reactions. It may also affect the operation of equipment, such as increasing the risk of short circuits in electrical equipment due to moisture, and indirectly affecting the normal mixing production. Changes in air pressure are related to the dissolution of gas in fracturing fluid to a certain extent, especially when it involves the use of processes containing gas components (such as carbon dioxide, nitrogen and other gases in foam fracturing fluid). Different air pressures will change the solubility of gas in the liquid, thereby affecting the overall performance and ratio requirements of the fracturing fluid.

[0047] Altitude. There are differences in environmental conditions such as atmospheric pressure and temperature in different altitudes. In addition, the air may be thin in high altitude areas, which will affect the operating efficiency of the equipment (such as motor power output, etc.) and the storage and transportation of raw materials, thereby affecting the mixing production process and the quality of the final fracturing fluid.

[0048] Geological conditions: If there are special circumstances in the geological environment of the oil field, such as leakage in the surrounding strata or the emission of special gases, these may affect the environmental atmosphere at the mixing site, change the storage conditions of the raw materials, or interact with the raw materials. Therefore, this is also one of the environmental factors that need to be considered.

[0049] Raw material temperature. Different raw materials have different initial temperatures when entering the mixing system, and their temperature will affect the physical properties of the raw materials themselves (such as density, viscosity, etc.) and chemical reactivity. For example, the heat transfer of high-temperature raw materials during the mixing process will change the temperature field of the entire mixing system, thereby affecting the interaction and mixing effect between the components, and affecting the concentration of the fracturing fluid. The ambient temperature interacts with the temperature of the mixing system. The ambient temperature will exchange heat with the mixing system through heat transfer. In the absence of effective temperature control measures, changes in ambient temperature may make it difficult for the mixing system to maintain an ideal temperature range, thereby affecting key parameters such as the reaction rate and solubility of the raw materials during the mixing process, and ultimately affecting the concentration and performance of the fracturing fluid.

[0050] Conductivity is a physical quantity that measures the electrical conductivity of a solution. For many raw material solutions (such as various ionic additive solutions in water-based fracturing fluids, etc.), their conductivity is closely related to the ion concentration in the solution. By testing the conductivity, the concentration and purity of the raw materials can be indirectly inferred. During the mixing process, the change in raw material conductivity can indicate whether the raw materials meet the requirements and whether the mixing has achieved the expected ion concentration balance, thereby assisting in determining whether the fracturing fluid concentration is accurate.

[0051] As different raw materials are mixed, the conductivity of the mixed system will change accordingly. This change pattern can be used as an important indicator to monitor whether the mixing process is proceeding normally in real time. If the conductivity change does not conform to the expected theoretical curve, it may mean that there is abnormal addition of raw materials or chemical reactions have occurred, which have affected the ion composition of the system. This helps to find problems in time and adjust the mixing operation.

[0052] The initial concentration of each raw material. Each component of the fracturing fluid has its own specified initial concentration requirements. For example, the concentration of the base polymer solution determines the key performance such as the thickening effect it can provide after mixing. The initial concentration of various additives (such as drag reducers, anti-swelling agents, etc.) is related to whether they can play the expected function in the fracturing fluid. Accurately grasping the initial concentration of each raw material is the basis for achieving precise mixing and ensuring that the final concentration of the fracturing fluid meets the construction requirements.

[0053] The impact of concentration fluctuations: The initial concentration of raw materials may fluctuate due to different raw material batches, changes in storage conditions, etc., and this fluctuation will be directly transmitted to the final mixed fracturing fluid, resulting in unstable concentration. Real-time monitoring and consideration of changes in initial concentration, and corresponding adjustments in automatic mixing control are crucial to maintaining the stability of fracturing fluid quality.

[0054] The input flow of each raw material enters the mixing system at a certain flow rate. The size of the flow rate directly determines the amount of each raw material added per unit time, which in turn affects the ratio between them. Accurately controlling the input flow of each raw material is a key operating link to achieve the preset mixing ratio. For example, the flow of water, polymer solution and various additive solutions needs to be accurately controlled according to the design requirements, and the accuracy of the fracturing fluid concentration is ensured by reasonable matching of the flow rate.

[0055] The liquid flow stability of the entire mixing system is also very important. Unstable flow may be caused by equipment failure (such as pump flow fluctuations) or pipeline blockage, leakage, etc., which not only affects the uniformity of the mixing, but may also cause some raw materials to stay in the system abnormally for a long time, affecting the chemical reaction process and the final concentration and performance of the fracturing fluid. Therefore, it is necessary to focus on the production data and conduct real-time adjustment and control.

[0056] In summary, comprehensive and detailed consideration of equipment status, environmental factors and these production parameters will help achieve more accurate and scientific control of the automatic mixing process of oilfield fracturing fluid, thereby ensuring that the quality and performance of the fracturing fluid meet the requirements of oilfield production operations.

[0057] As an optional embodiment, it is assumed that the concentration prediction model includes at least: an extraction layer, a construction layer, an association layer, and a prediction layer. Based on this, in 102, the concentration prediction model is used to perform real-time prediction on the mixed production data to obtain the concentration prediction value of the oilfield fracturing fluid, which can be implemented as follows:

[0058] 201, extracting a compounding production feature from the compounding production data through the extraction layer; the type of the compounding production feature is dynamically configured based on the production environment and / or production demand;

[0059] 202, constructing the mixed production characteristics into corresponding production element nodes in the oilfield fracturing fluid association graph network through the construction layer; the production element nodes at least include: equipment status nodes, environmental factor nodes, and production parameter nodes;

[0060] 203, obtaining the correlation relationship between the production element nodes through the correlation layer, and constructing a network connection structure between the production element nodes based on the correlation relationship; the correlation relationship at least includes: the spatiotemporal correlation weight, the ratio concentration correlation, and the degree of influence on the ratio result between the production element nodes;

[0061] 204 , predicting the ratio result according to the oilfield fracturing fluid association graph network through the prediction layer to obtain a concentration prediction value of the oilfield fracturing fluid.

[0062] In the present application embodiment, the concentration prediction model is a neural network specially used for processing graph structure data, which can effectively capture the relationship between the topological structure information and nodes in the data. For example, the model structure of the graph convolutional network (Graph Convolutional Network, GCN), the graph attention network (Graph Attention Network, GAT) etc. can be used as the basis to build a concentration prediction model. It performs well when processing data with complex associations. For example, for an oilfield fracturing fluid system, if various influencing factors and monitoring data are constructed into a graph structure, the potential relationship therein can be better excavated, thereby improving the accuracy of concentration prediction, and providing a more comprehensive basis for feedback regulation. If the relevant parameters, equipment status, environmental factors, etc. of the oilfield fracturing fluid are used as the nodes of the graph, the interaction between the nodes is used as an edge, and the complex relationship between these factors can be captured, and then the concentration of the fracturing fluid can be more accurately predicted, and a more reasonable strategy can be provided for feedback regulation.

[0063] In 201, the mixed production features are extracted from the mixed production data through the extraction layer. The extraction layer will dynamically determine the type of features to be extracted based on the specific production environment of the oilfield fracturing fluid (such as different oilfield geographical locations, climatic conditions, etc.) and actual production needs (such as the requirements of a specific fracturing process for fracturing fluid performance, etc.). For example, in an oilfield under a high temperature environment, more emphasis may be placed on extracting raw material characteristics related to temperature sensitivity; for production needs that require high-viscosity fracturing fluids, the focus will be on extracting features such as the concentration of key components that affect viscosity.

[0064] For different types of data (equipment status data may be numerical or status-coded, environmental factor data may include continuous meteorological data, and production parameter data may also be in multiple formats), corresponding data processing and feature extraction techniques are used. For example, for time series data in equipment status (such as historical records of speed changes in delivery pumps), time series analysis methods can be used to extract statistical features such as mean, variance, and autocorrelation coefficient; for image data in environmental factors (if there are camera images that monitor the on-site environment), convolutional neural networks and other methods can be used to extract texture, color and other features to reflect environmental conditions, thereby converting the original mixed production data into mixed production features that are valuable for subsequent predictions. In this way, key features closely related to the concentration of fracturing fluid can be screened out from massive and complex mixed production data, redundant and irrelevant information can be removed, data dimensions can be reduced, and subsequent models can be prevented from processing too much useless data, improving data processing efficiency and model operation speed, while allowing the model to focus more on learning the laws of factors that truly affect concentration.

[0065] Specifically, feature types are dynamically configured based on the production environment and demand, making the system highly adaptable. Regardless of changes in the production environment (such as seasonal changes leading to drastic temperature changes, new oil field extraction sites with special geological conditions, etc.) or adjustments in production demand (such as improved fracturing technology that places new requirements on fracturing fluid performance), the system can extract appropriate features in a timely manner to ensure that the concentration prediction model is always learning and predicting based on the data that best fits the actual situation.

[0066] In 202, the mixed production feature is constructed as a corresponding production element node in the oilfield fracturing fluid association graph network through the construction layer. Specifically, the construction layer maps the production feature to different types of production element nodes, namely, equipment status nodes, environmental factor nodes, production parameter nodes, etc., according to the nature and category of the production feature. For example, the equipment-related features such as the rotation speed and valve opening of the delivery pump are integrated and constructed as equipment status nodes, the environmental data features such as temperature and humidity are corresponding to the environmental factor nodes, and the production parameter features such as the concentration and flow rate of various raw materials are constructed as production parameter nodes. Each node represents a specific aspect of the production factor and carries the characteristic information related to it. Each production element node is given a corresponding attribute, and the attribute value is the corresponding characteristic value extracted from the mixed production data. In this way, the abstract data feature is visualized as a node entity in the graph network, so that the relationship between the elements can be analyzed later using the structure of the graph.

[0067] In this way, the dispersed and mixed production characteristics of different dimensions are structured and organized in the form of graph network nodes, which clearly presents the division and distribution of different elements in the production process, facilitates intuitive understanding of the position and role of each element in the entire production system, and also lays the foundation for further exploring the complex relationship between them. This method of constructing nodes according to equipment, environment, production parameters, etc. conforms to the logical architecture of the actual production process of oilfield fracturing fluid, and can better reflect the mutual influence and mutual restriction relationship between various production factors in reality, making the subsequent analysis and prediction based on the graph network closer to the actual production scenario, and improving the model's ability to characterize the actual production situation and the accuracy of prediction.

[0068] In 203, the association relationship between the production element nodes is obtained through the association layer, and a network connection structure between the production element nodes is constructed based on the association relationship.

[0069] Considering the continuity and correlation of the production process in time and space, the time-space correlation weights between the nodes of each production element are determined by analyzing historical data and real-time monitoring data, and using appropriate mathematical methods (such as calculating the correlation between nodes at different times based on time series analysis, and using spatial distance measurement or topological relationship to determine the degree of spatial correlation). For example, the changes of a certain equipment status node (such as the status of a delivery pump) at different time points may be temporally correlated with subsequent environmental factor nodes (such as temperature changes after a period of time, because the operation of the equipment may generate heat and affect the ambient temperature) and production parameter nodes (such as changes in raw material flow, which are affected by the equipment's transportation capacity); in space, the state changes of equipment at adjacent locations may affect each other, or the environmental factors in different regions (such as temperature differences at different pipe network nodes) may have spatial correlations on the production parameters. The time-space correlation weights are obtained by quantifying these correlations.

[0070] It is understandable that the intrinsic relationship between the characteristics represented by each production element node and the final fracturing fluid ratio concentration is deeply studied, and the degree of correlation between them is calculated through statistical analysis, machine learning regression and other methods to determine the correlation between the ratio concentration. For example, the direct relationship between the initial concentration node of the raw material and the final fracturing fluid concentration, or the indirect relationship between the ambient temperature node and the ratio concentration by affecting the properties of the raw materials, etc., is analyzed to find out the direction and intensity of the influence of each node on the concentration.

[0071] Specifically, appropriate impact analysis methods (such as feature importance evaluation algorithms, causal inference methods, etc.) are used to measure the degree of influence of each production element node on the final ratio result (i.e., fracturing fluid concentration) and determine the relative importance weights of different nodes in the entire production system. For example, by comparing the impact of changing the value of a certain equipment status node on the final concentration change when other conditions remain unchanged, or analyzing the degree of influence of environmental factor nodes on the reaction process of different raw materials and thus on the concentration, etc., the quantitative value of the degree of influence of each node is obtained.

[0072] Based on the above-mentioned various correlation relationships, according to the set threshold or weight standard, when the correlation weight between two production element nodes reaches a certain level, a network connection is established between them, thereby forming a connection structure of the oilfield fracturing fluid correlation graph network, and the relationship between the nodes is represented by the edges of the graph.

[0073] Further optionally, in 203, the production element node v is obtained in the association layer i and produce element node v j The spatiotemporal correlation weight W ts The process is expressed as the following expression:

[0074]

[0075] in, The square of the standard deviation parameter representing the time correlation, It is used to control the magnitude of the weight change with time delay. t represents the time delay factor corresponding to the current device and production environment. ij Indicates the production element node v in the mixed production data i and produce element node v j The average time delay of the association, α represents the production element node v i and produce element node v j The time correlation weight ratio between Represents the production element node v i and produce element node v j The time correlation weight between them, β represents the production element node v i and produce element node v j The spatial correlation weight ratio between Represents the production element node v in the pipe network system of the actual production equipment i and produce element node v j The spatial correlation weight between ||v i -v j ||2 represents the production element node v in the pipe network system i and produce element node v j The equivalent spatial distance between .

[0076] In another embodiment, the production element node v i The set of eigenvalues ​​associated with the ratio concentration is c i , produce element node v j The set of eigenvalues ​​associated with the ratio concentration is c j , the characteristic value set c is obtained in the association layer i and the eigenvalue set c j The combined ratio concentration correlation evaluation value W c The process is expressed as the following expression:

[0077]

[0079] Among them, c i t and c j t Respectively represent the eigenvalue set c i and the eigenvalue set c j The eigenvalue in the sample at the tth moment, and Respectively represent the eigenvalue set ci and the eigenvalue set c j The respective sample means, T is the total number of samples.

[0080] In another embodiment, the step of obtaining the influence degree of the ratio result of the production element node in the association layer can be implemented as follows: taking the characteristic value of each production element node in the historical mixed production data as an independent variable and taking the historical fracturing fluid concentration corresponding to each production element node as the target variable, constructing a multivariate linear regression analysis model; i Input into the multiple linear regression analysis model to obtain the production element node v i The regression coefficient vector of each eigenvalue in; the production element node v i The absolute value of the regression coefficient vector of each eigenvalue in is normalized to obtain the production element node v i The corresponding impact assessment value of the matching result.

[0081] For example, in the process of oil field exploitation, the ratio of hydraulic fluid (also called fracturing fluid) has an important influence on the fracturing effect of oil wells. Hydraulic fluid is usually composed of a variety of chemical additives (such as thickeners, cross-linking agents, degelling agents, etc.). In order to optimize the ratio of hydraulic fluid, a multivariate linear regression analysis model can be used to evaluate the degree of influence of different additives on the performance of fracturing fluid. Collect hydraulic fluid performance data (such as viscosity, stability, degelling time, etc.) at different concentrations of different additives over a period of time. The concentration of each additive (such as thickener A, cross-linking agent B, degelling agent C). The target variable is the historical hydraulic fluid concentration (such as viscosity, stability, etc.).

[0082] Further optionally, in 203, a network connection structure between the production element nodes is constructed based on the association relationship. Specifically, an optional method is to use the Gini impurity evaluation method to evaluate the production element node v i The spatiotemporal correlation weight, the ratio concentration correlation evaluation value, and the evaluation value of the degree of influence on the ratio result are comprehensively optimized; the spatiotemporal correlation weight, the ratio concentration correlation evaluation value, and the evaluation value of the degree of influence on the ratio result after comprehensive optimization are weight parameter mapped to obtain the production element node v i Connection parameters between the surrounding nodes; Setting and constructing the production element node v based on the connection parameters i The associated topology.

[0083] For example, in the process of oilfield exploitation, by constructing a network connection structure between the nodes of hydraulic fluid production elements, the Gini impurity evaluation method is used to comprehensively consider the temporal and spatial correlation weights of the additives, the ratio concentration correlation evaluation value and the degree of influence on the ratio results, so as to optimize the ratio of the hydraulic fluid. Taking thickener A, cross-linking agent B, and degelling agent C as examples, the temporal and spatial correlation weights are analyzed through historical data, the concentration influence of each additive is evaluated using a multivariate linear regression model, and the regression coefficient is normalized to obtain the impact degree evaluation value. The impact of the changes in additives at different time points on the performance of the hydraulic fluid is analyzed through historical data. The multivariate linear regression analysis model is used to evaluate the impact of the concentration of each additive on the performance of the hydraulic fluid. The absolute value of the regression coefficient vector is used for normalization to obtain the impact degree evaluation value of each additive. The temporal and spatial correlation weights, the ratio concentration correlation evaluation value and the impact degree evaluation value on the ratio result are weighted and integrated to obtain a comprehensive evaluation value. Assuming that the weight parameters α=0.4, β=0.3, γ=0.3 are set, the comprehensive evaluation value is calculated and mapped to the connection parameters between nodes, and finally the optimized topological structure between thickener A, crosslinker B, and degumming agent C is constructed, and the connection parameters are 0.29, 0.47, and 0.34 respectively, thereby improving the accuracy and stability of the hydraulic fluid ratio. For example, the connection strength between the production element nodes is set according to the connection parameters. In this way, an optimized network connection structure between the production element nodes can be constructed, so as to better understand and optimize the hydraulic fluid ratio process.

[0084] In this way, the complex correlations hidden between various factors in the production process are fully and deeply excavated. Instead of looking at various factors such as equipment, environment, and production parameters in isolation, it reveals how they interact and affect each other on the final fracturing fluid concentration from a system perspective, providing a more comprehensive and detailed basis for accurate prediction of concentration. The network connection structure is constructed based on the accurately calculated correlation relationship, so that the graph network can truly reflect the actual production situation. Each connection represents a meaningful production factor association, avoiding the problem of model inaccuracy caused by arbitrarily constructed networks. In addition, when information is subsequently transmitted and analyzed based on this reasonably constructed network, the transmission path and synergistic relationship of key factors affecting concentration can be more effectively captured, thereby improving the accuracy of concentration prediction.

[0085] In 204, the prediction layer predicts the ratio result according to the oilfield fracturing fluid association graph network to obtain the concentration prediction value of the oilfield fracturing fluid. The prediction layer uses the message passing mechanism of the graph neural network (such as the graph convolution network GCN as an example) to perform information propagation and aggregation operations on the constructed oilfield fracturing fluid association graph network. The characteristic information of each production element node will be transmitted and fused between adjacent nodes along the network connection (edge). After multiple layers of iterative updates, each node continuously absorbs information from neighboring nodes and its own historical state, and gradually integrates the global information in the entire graph network, so as to realize the continuous optimization and update of the feature representation of each node, so that it contains more abundant information related to the ratio result. After multi-layer information processing of the graph neural network, the node state vector obtained in the last layer is input into the fully connected layer. The fully connected layer linearly transforms and combines the node information through the learned weight parameters, and adds the bias term to finally output the concentration prediction value of the oilfield fracturing fluid, and converts the association relationship of each element contained in the graph network into a specific concentration prediction value.

[0086] In this way, by using the complex correlations between the nodes of each production element captured by the graph network for prediction, the synergistic and indirect effects between different elements can be fully considered. Compared with the traditional prediction method based on simple linear relationships or isolated features, the final fracturing fluid concentration can be more accurately inferred based on the current production situation, improving the accuracy and reliability of the prediction. Through the information propagation and aggregation process of the graph neural network, the effective integration of the global information of the entire production system is achieved, so that the prediction results do not only rely on a few local factors, but also integrate the overall consideration of all relevant production factors and their interrelationships, better adapt to the complex and changing production environment and the dynamic changes in the production process, and help to generate more realistic concentration prediction values, providing strong support for subsequent automatic mixing control.

[0087] In summary, these four steps improve the accuracy, adaptability and fit of oilfield fracturing fluid concentration prediction to actual production conditions in many aspects by constructing an association graph network based on production data characteristics and using it to predict concentration, which is of great significance for achieving precise automatic mixing control.

[0088] Further optionally, in 204, the prediction layer predicts the ratio result according to the oilfield fracturing fluid association graph network to obtain the concentration prediction value of the oilfield fracturing fluid, which can be implemented as follows:

[0089] According to the correlation relationship between the production element nodes determined in the oilfield fracturing fluid association graph network, the corresponding weight matrix is ​​calculated; the weight matrix is ​​converted into an adjacency matrix, and the oilfield fracturing fluid association graph network is subjected to information propagation and aggregation operations, and the adjacency matrix is ​​processed to obtain the correlation matrix of the oilfield fracturing fluid association graph network; the correlation matrix at least includes a unit matrix and a degree matrix matched with the adjacency matrix; the weight matrix is ​​optimized based on the correlation matrix; the node feature matrix corresponding to the oilfield fracturing fluid association graph network is constructed using the optimized weight matrix; the node feature matrix is ​​iteratively updated and the full connection layer is calculated to obtain the concentration prediction value of the oilfield fracturing fluid.

[0090] For example, assuming there is an oilfield fracturing fluid production scenario, the equipment status node contains three features: "delivery pump A speed", "delivery pump B speed" and "agitator power", and their value ranges are [1000, 3000] rpm, [800, 2500] rpm and [5, 20] kW. The environmental factor node covers two features: "ambient temperature" and "ambient humidity", and the value ranges are [10°C, 35°C] and [30%, 80%]. The production parameter node has three features: "raw material X concentration", "raw material Y concentration" and "total liquid flow", and the corresponding value ranges are [0.1, 0.5] g / L, [0.2, 0.6] g / L and [10, 50] m3 / h. Further optionally, a large amount of historical production data within a certain period of time is collected, and the characteristic values ​​of these production element nodes at different times and the final actual fracturing fluid concentration are recorded. Now, an association graph network is constructed based on these data and concentration prediction is performed.

[0091] According to the correlation relationship between each production element node determined in the oilfield fracturing fluid association graph network, the corresponding weight matrix is ​​calculated. By analyzing the historical production data, the correlation relationship between each production element node is calculated. For example, it is found that there is a strong positive correlation between "delivery pump A speed" and "liquid total flow" (because the delivery pump speed directly affects the raw material delivery amount, and then affects the total flow), "ambient temperature" and "raw material X concentration" have a certain negative correlation (temperature may affect the solubility of the raw material, etc., resulting in concentration changes), etc. Quantifying these relationships into weight values ​​in the weight matrix can comprehensively and accurately dig out the complex mutual influence relationship between different production factors, no longer look at each factor in isolation, but understand how they work together from the perspective of the overall system to affect the final fracturing fluid concentration. Among them, the weight matrix provides an accurate quantitative basis for the subsequent construction of the connection structure of the association graph network, so that the network can truly reflect the close connection between the various elements in the actual production process, ensure that the network structure fits the actual production logic, and lay the foundation for subsequent accurate concentration prediction.

[0092] For example, after analysis and calculation, the correlation weight between "speed of delivery pump A" and "total liquid flow" is 0.8 (only an example, actually obtained based on statistical analysis of data), indicating that the two are highly correlated, and this weight will be used to reflect the close connection between them when the network is subsequently constructed.

[0093] Furthermore, the weight matrix is ​​converted into an adjacency matrix, and the oilfield fracturing fluid association graph network is subjected to information propagation and aggregation operations, and the adjacency matrix is ​​processed to obtain a correlation matrix of the oilfield fracturing fluid association graph network; the correlation matrix at least includes a unit matrix and a degree matrix that match the adjacency matrix.

[0094] In this way, the process of converting the weight matrix into the adjacency matrix is ​​actually to determine whether the nodes in the network are connected according to the strength of the correlation between each node, and to construct the topological structure of the oilfield fracturing fluid association graph network. For example, a weight threshold is set to 0.5. When the correlation weight between two nodes is greater than 0.5, the corresponding element value in the adjacency matrix is ​​the weight value, indicating that there is a connection, otherwise it is 0, indicating that there is no connection. The network structure constructed in this way can intuitively present the truly influential association paths between the production elements, so that the entire graph network can concisely and accurately reflect the relationship between production factors, and avoid irrelevant or weakly correlated factors from interfering with subsequent analysis and prediction. In the information dissemination and aggregation operation, the adjacency matrix is ​​processed using the correlation matrix (the unit matrix ensures that the node's own information participates in the transmission, and the degree matrix is ​​used to normalize the weight of information transmission) so that the information between nodes can flow and merge in the graph network according to reasonable rules. For example, a node can receive information from its connected neighbor nodes and integrate and update the information based on the importance of itself and its neighbor nodes (reflected by the degree matrix), so that each node can continuously absorb global information in the graph network, enrich and optimize the feature representation of each node, and prepare for accurate concentration prediction.

[0095] Assuming that in the constructed adjacency matrix, the element value corresponding to the "Speed ​​of Pump A" node and the "Total Liquid Flow" node is 0.8 (indicating connection and a weight of 0.8), in the process of information propagation and aggregation, the information of the "Speed ​​of Pump A" node will be transmitted to the "Total Liquid Flow" node according to the corresponding rules (based on the method processed by the correlation matrix) and participate in its feature update. At the same time, the information of the "Total Liquid Flow" node will also be transmitted in the opposite direction, so that the feature representations of the two nodes are integrated with the relevant information of each other, reflecting the actual production status more comprehensively.

[0096] In this way, the correlation matrix incorporates the overall topological information of the nodes themselves and their connections. By using this information to optimize the weight matrix, the accuracy of the correlation weights between nodes can be further corrected. For example, the weight between two nodes that was initially calculated may not fully take into account their position in the entire network structure and the influence of surrounding nodes on them. By combining the information of the correlation matrix, this weight can be reasonably adjusted to make it more consistent with the actual situation of the mutual influence of various factors in actual production, making subsequent weight-based analysis and prediction more reliable.

[0097] As the production process continues and new data is continuously collected, the graph network structure may change to a certain extent (such as the emergence of new correlations or changes in the strength of existing relationships). The weight optimization mechanism based on the correlation matrix can adapt to this dynamic change in a timely manner and dynamically adjust the weight value to ensure that the graph network always accurately reflects the latest correlation status of production factors, thereby ensuring the effectiveness and accuracy of the concentration prediction model.

[0098] For example, as the production environment temperature changes seasonally, the original correlation weight between "ambient temperature" and "raw material X concentration" may need to be optimized and adjusted based on new data and changes in the network structure (other related node status changes and other factors), so that the weight is more in line with the current actual production situation and the accuracy of temperature considerations in concentration prediction is improved.

[0099] Next, the node feature matrix corresponding to the oilfield fracturing fluid association graph network is constructed using the optimized weight matrix. The node feature matrix integrates the characteristic values ​​of each production element node with the optimized weight information. Each node not only contains its own original characteristic data, but also incorporates the weight information associated with other nodes, so that the information carried by the node is more comprehensive and has practical significance. For example, the representation of the "raw material X concentration" node in the node feature matrix is ​​no longer just a concentration value, but also combines its associated weight with other related nodes (such as the weight situation affected by the "ambient temperature", "delivery pump A speed" and other nodes), so that the influence of the synergistic effect of various factors on the concentration can be more fully reflected in the subsequent calculation and analysis. The constructed node feature matrix provides a suitable data structure foundation for the subsequent iterative update operation, which is convenient for performing multiple information transmission, feature fusion and other operations according to the established graph neural network algorithm (such as based on graph convolution network, etc.), and gradually digging out the hidden relationship between deeper production factors, thereby improving the accuracy of the final concentration prediction.

[0100] Then, the node feature matrix is ​​iteratively updated and the fully connected layer is calculated to obtain the concentration prediction value of the oilfield fracturing fluid. In this way, through the iterative update operation, the information of each node in the node feature matrix will continuously absorb and integrate the information transmitted by other nodes in the graph network as the number of iterations increases, making full use of the dynamic change information of each element in the production process and the complex relationship between them. For example, after many iterations, the information of the "equipment status node" may integrate the influence of various aspects such as changes in environmental factors and adjustments in production parameters, so that its feature representation is more in line with the comprehensive status of the current actual production, and then when predicting the concentration, it can more comprehensively consider the influence of various real-time changing factors, thereby improving the accuracy and real-time performance of the prediction.

[0101] The fully connected layer performs calculations based on the node feature matrix that has been enriched and optimized after multiple iterations, and finally converts the production factors and their associated relationship information integrated in the graph network into specific concentration prediction values. This method of combining the graph network feature representation with the mapping of the fully connected layer can accurately output the concentration prediction results that meet the requirements according to the actual production situation, provide strong data support for the subsequent automatic mixing control, help to achieve accurate raw material allocation, produce fracturing fluid that meets quality requirements, and improve the success rate and efficiency of the entire oilfield fracturing operation.

[0102] For example, the final concentration prediction value calculated by the full connection layer is 2.2 g / L (assuming the target concentration range is 2.0-2.5 g / L). Based on this prediction value, the raw material ratio and other operations can be adjusted in time to ensure that the actual mixed fracturing fluid concentration is within a reasonable range and meets the requirements of oilfield production and construction for fracturing fluid.

[0103] In summary, through the above steps and their corresponding beneficial technical effects, the concentration prediction using the oilfield fracturing fluid association graph network can better adapt to the complex and changeable production environment and the dynamic relationship between various production factors, improve the accuracy and reliability of concentration prediction, and play an important and positive role in the automatic mixing control of oilfield fracturing fluid.

[0104] As an optional embodiment, in 103, real-time proportioning information of the oilfield fracturing fluid is generated based on the concentration prediction value, and automatic mixing control of the oilfield fracturing fluid is performed through the real-time proportioning information.

[0105] In the embodiment of the present application, a self-organizing recurrent fuzzy neural network (SORFNN) can be used. In this way, the advantages of self-organization, recursion and fuzzy logic are combined. It can automatically adjust the network structure according to the characteristics of the input data, while using the recursive structure to process the dynamic information in the time series data, and model uncertainty and ambiguity through fuzzy logic.

[0106] The prediction effect is better for data with nonlinear dynamic changes. For example, in the prediction of oilfield fracturing fluid concentration, it can screen out characteristic variables with strong correlation, automatically adjust the network structure, and have high prediction accuracy. For oilfield fracturing fluid concentration, which is affected by multiple factors and changes dynamically, SORFNN can better adapt to the complexity of data and improve the accuracy and reliability of prediction. It can be used to monitor and analyze the changing trend of fracturing fluid concentration in real time, automatically adjust the prediction model according to historical data and current input, and provide timely and accurate concentration prediction information for feedback regulation to achieve precise control of the fracturing fluid mixing process.

[0107] For example, suppose that a mixed production operation of oilfield fracturing fluid is being carried out, involving three main raw materials: water, polymer (for adjusting viscosity) and an additive (with anti-swelling and other functions). Their ratios must be accurately controlled according to the real-time situation to produce a fracturing fluid that meets the requirements, and it is known that the concentration of the fracturing fluid is affected by multiple factors such as ambient temperature, equipment operation time, and raw material batch differences, and these factors are in dynamic change. Based on this, collect detailed data for each mixing production over a period of time (for example, the past month), including ambient temperature (recorded once every hour, ranging from 5°C to 30°C), equipment operation time (starting from the start of the equipment, each mixing is recorded at the end, ranging from 1 to 8 hours), the initial concentration of each raw material (the purity of water is basically stable, but the concentration of polymer and additive has certain fluctuations, the polymer concentration range is 0.1-0.3g / L, the additive concentration range is 0.05-0.15g / L), the input flow rate of each raw material (water flow range is 10-30m 3 / h, polymer flow range 1-5m 3 / h, additive flow range 0.5-2m 3 / h) and the actual measured fracturing fluid concentration value (ranging from 1.2-2.5g / L). These data constitute the basic data set for the subsequent construction of the SORFNN model and can be regarded as data with time series characteristics, because each mixing production is recorded in chronological order. At the same time, the real-time flow rate, ambient temperature, equipment operation time and other data of each raw material currently being mixed are collected in real time. These real-time data will be used as input for real-time concentration prediction and ratio adjustment.

[0108] First, determine the initial structure of the SORFNN model, and set the number of input layer nodes to correspond to the number of input data features. In this example, there are 7 features, such as ambient temperature, equipment operation time, initial concentration of each raw material, and input flow rate of each raw material, so the number of input layer nodes is set to 7. The intermediate fuzzy rule layer, recursive layer, etc. are set based on experience or some simple tentative settings to determine the approximate number of nodes and connection methods. For example, the fuzzy rule layer is initially set to 10 nodes, and the recursive layer is set to 5 nodes. The number of output layer nodes is set to 1, corresponding to the predicted fracturing fluid concentration value. The preprocessed historical data set is divided according to a certain ratio (such as 70% as a training set, 20% as a validation set, and 10% as a test set), and the training set data is input into the SORFNN model for training. During the training process, the self-organizing characteristics of SORFNN begin to play a role. For example, it will automatically adjust the number of nodes in the fuzzy rule layer and the corresponding fuzzy rule parameters according to the correlation between the input data features and the degree of influence on the output (fracturing fluid concentration). Assuming that as training progresses, it is found that the two characteristics of ambient temperature and additive concentration have a specific nonlinear relationship in affecting concentration, SORFNN will automatically generate or adjust the corresponding fuzzy rules to better describe this relationship, so that the network structure can adapt to the characteristics of the data.

[0109] At the same time, the recursive layer uses the time series characteristics of the data to process the dynamic effects of time-varying factors such as equipment operating time on concentration. For example, the longer the equipment runs, the more likely it is that slight changes in equipment performance will affect the mixing effect of the raw materials and thus affect the concentration. The recursive structure can capture this dynamic change information accumulated over time and feed it back to the network for concentration prediction. Through continuous iterative training (using appropriate training algorithms, such as the gradient descent algorithm combined with back propagation to adjust the weight parameters of each layer of the network), and adjusting the model hyperparameters (such as learning rate, number of iterations, etc.) according to the prediction error on the validation set (such as evaluation indicators such as mean square error), overfitting is prevented and the model is continuously optimized until satisfactory prediction accuracy is obtained on the test set.

[0110] In the actual mixing production process, the current ambient temperature, equipment operation time, and real-time flow rate and concentration of each raw material are collected and preprocessed in real time and input into the trained SORFNN model. The model uses its learned structure and parameters to quickly predict the concentration. For example, the current ambient temperature is 15°C (the normalized value is the corresponding calculation result), the equipment has been running for 3 hours, and the flow rates of each raw material are 20m3 / s of water, 20m4 / s of water, and 20m5 / s of water. 3 / h, polymer 3m 3 / h, additive 1m 3 / h and other conditions, the model outputs a predicted fracturing fluid concentration value, which is assumed to be 1.8g / L. Real-time ratio information is generated based on the pre-set target concentration range and the current predicted concentration value. Assuming that the target concentration requirement is 2.0g / L, and the current predicted concentration is 1.8g / L, it means that the concentration is too low and the raw material ratio needs to be adjusted to increase the concentration. The SORFNN model combines the relationship between various factors and concentration learned internally (for example, knowing that increasing the polymer concentration can effectively increase the concentration, and the approximate adjustment range under the current ambient temperature, equipment status and other conditions), and automatically generates adjustment suggestions, such as increasing the polymer flow rate from the current 3m 3 / h is appropriately increased to 3.5m 3 / h, and can also fine-tune the flow of additives, etc., to form new real-time ratio information, guide the real-time addition of each raw material, so as to achieve the concentration of the fracturing fluid close to the target value. According to the generated real-time ratio information, the control system automatically adjusts the speed or valve opening of each raw material delivery pump to accurately change the input flow of each raw material and achieve mixing according to the new ratio. For example, the control system sends a command to the polymer delivery pump to increase its speed so that the polymer flow rate is increased to the recommended 3.5m / h. 3 / h, and adjust the opening of the additive valve at the same time so that the additive flow rate also reaches the adjusted value, thereby ensuring that the entire mixing process can be dynamically adjusted in real time according to the concentration prediction, accurately controlling the concentration of the fracturing fluid, making it as stable as possible within the target concentration range, and meeting the quality requirements of the fracturing fluid in oilfield fracturing operations. During the entire mixing production process, the SORFNN model will continue to receive real-time data, continuously update the concentration prediction results and adjust the real-time ratio information accordingly, to achieve cyclic, dynamic and precise control of the oilfield fracturing fluid mixing process, effectively respond to the dynamic changes of various factors, and improve the mixing quality and efficiency.

[0111] In the embodiment of the present application, first, the mixed production data of the oilfield fracturing fluid is obtained; the mixed production data at least includes: equipment status, environmental factors, production parameters; the production parameters at least include: temperature, conductivity, initial concentration, liquid flow rate. Then, the mixed production data is predicted in real time through the concentration prediction model to obtain the concentration prediction value of the oilfield fracturing fluid; the concentration prediction model is trained using the actual concentration value of the fracturing fluid obtained under different production conditions; the actual concentration value of the fracturing fluid comes from the detection data of multiple pipeline nodes in the production environment. Finally, the real-time proportioning information of the oilfield fracturing fluid is generated based on the concentration prediction value, and the automatic mixing control of the oilfield fracturing fluid is performed through the real-time proportioning information. In summary, the automatic mixing control method of the oilfield fracturing fluid helps to improve the accuracy, efficiency and quality of the fracturing fluid mixing through many advantages, enhances the adaptability to complex production environments, and promotes the intelligent development of production management, which has positive and important significance for oilfield mining operations.

[0112] In another embodiment of the present application, an automatic mixing control device for oilfield fracturing fluid is also provided. Figure 3 As shown, the device includes the following units:

[0113] An acquisition unit is configured to acquire mixed production data of oilfield fracturing fluid; the mixed production data at least includes: equipment status, environmental factors, production parameters; the production parameters at least include: temperature, conductivity, initial concentration, liquid flow rate;

[0114] The prediction unit is configured to perform real-time prediction on the mixed production data through a concentration prediction model to obtain a concentration prediction value of the oilfield fracturing fluid; the concentration prediction model is trained using actual concentration values ​​of the fracturing fluid obtained under different production conditions; the actual concentration value of the fracturing fluid comes from detection data of multiple pipe network nodes in the production environment;

[0115] The execution unit is configured to generate real-time ratio information of the oilfield fracturing fluid based on the concentration prediction value, and execute automatic mixing control of the oilfield fracturing fluid through the real-time ratio information.

[0116] Further optionally, the concentration prediction model at least includes: an extraction layer, a construction layer, an association layer, and a prediction layer; the prediction unit, through the concentration prediction model, performs real-time prediction on the mixed production data to obtain the concentration prediction value of the oilfield fracturing fluid, and is specifically configured as follows: through the extraction layer, extracting mixed production characteristics from the mixed production data; the type of the mixed production characteristics is dynamically configured based on the production environment and / or production demand; through the construction layer, constructing the mixed production characteristics as corresponding production element nodes in the oilfield fracturing fluid association graph network; the production element nodes at least include: equipment status nodes, environmental factor nodes, and production parameter nodes; through the association layer, obtaining the correlation relationship between the production element nodes, and constructing the network connection structure between the production element nodes based on the correlation relationship; the correlation relationship at least includes: the spatiotemporal correlation weights between the production element nodes, the ratio concentration correlation, and the degree of influence on the ratio result; through the prediction layer, predicting the ratio result according to the oilfield fracturing fluid association graph network to obtain the concentration prediction value of the oilfield fracturing fluid.

[0117] Further optionally, the production element node v is obtained in the association layer i and produce element node v j The spatiotemporal correlation weight W ts The process is expressed as the following expression:

[0118]

[0119] in, The square of the standard deviation parameter representing the time correlation, It is used to control the magnitude of the weight change with time delay. t represents the time delay factor corresponding to the current device and production environment. ij Indicates the production element node v in the mixed production data i and produce element node v j The average time delay of the association, α represents the production element node v i and produce element node v j The time correlation weight ratio between Represents the production element node v i and produce element node v j The time correlation weight between them, β represents the production element node v i and produce element node v j The spatial correlation weight ratio between Represents the production element node v in the pipe network system of the actual production equipment i and produce element node v j The spatial correlation weight between ||v i -vj ||2 represents the production element node v in the pipe network system i and produce element node v j The equivalent spatial distance between .

[0120] Optionally, the element node v is generated i The set of eigenvalues ​​associated with the ratio concentration is c i , produce element node v j The set of eigenvalues ​​associated with the ratio concentration is c j , the characteristic value set c is obtained in the association layer i and the eigenvalue set c j The combined ratio concentration correlation evaluation value W c The process is expressed as the following expression:

[0121]

[0123] Among them, c i t and c j t Respectively represent the eigenvalue set c i and the eigenvalue set c j The eigenvalue in the sample at the tth moment, and Respectively represent the eigenvalue set c i and the eigenvalue set c j The respective sample means, T is the total number of samples.

[0124] Further optionally, the prediction unit, in the associated layer, obtains the influence degree of the ratio result of the production element node, and is configured as follows:

[0125] A multivariate linear regression analysis model was constructed with the characteristic values ​​of each production element node in the historical mixed production data as the independent variable and the historical fracturing fluid concentration corresponding to each production element node as the target variable.

[0126] Will produce element node v i Input into the multiple linear regression analysis model to obtain the production element node v i The regression coefficient vector of each eigenvalue in ;

[0127] Will produce element node v i The absolute value of the regression coefficient vector of each eigenvalue in is normalized to obtain the production element node v i The corresponding impact assessment value of the matching result.

[0128] Further optionally, the prediction unit constructs a network connection structure between the production element nodes based on the association relationship, and is specifically configured as follows:

[0129] The Gini impurity evaluation method is used to evaluate the production element node v i Comprehensively optimize the spatiotemporal correlation weight, the ratio concentration correlation evaluation value, and the evaluation value of the impact degree on the ratio result;

[0130] The optimized spatiotemporal correlation weight, the ratio concentration correlation evaluation value, and the evaluation value of the impact degree on the ratio result are mapped to weight parameters to obtain the production element node v i Connection parameters with surrounding nodes;

[0131] Set and construct the production element node v based on the connection parameters i The associated topology.

[0132] Further optionally, the prediction unit predicts the ratio result according to the oilfield fracturing fluid association graph network through the prediction layer to obtain the concentration prediction value of the oilfield fracturing fluid, which is specifically configured as follows:

[0133] Determine the correlation relationship between the production element nodes in the oilfield fracturing fluid correlation graph network, and calculate the corresponding weight matrix;

[0134] The weight matrix is ​​converted into an adjacency matrix, and the oilfield fracturing fluid association graph network is subjected to information propagation and aggregation operations, and the adjacency matrix is ​​processed to obtain a correlation matrix of the oilfield fracturing fluid association graph network; the correlation matrix at least includes a unit matrix and a degree matrix matched with the adjacency matrix;

[0135] Optimizing the weight matrix based on the correlation matrix;

[0136] Using the optimized weight matrix to construct a node feature matrix corresponding to the oilfield fracturing fluid association graph network;

[0137] The node feature matrix is ​​iteratively updated and fully connected layer calculated to obtain a concentration prediction value of the oilfield fracturing fluid.

[0138] In the embodiments of the present application, the automatic mixing control device for oilfield fracturing fluid helps to improve the accuracy, efficiency and quality of fracturing fluid mixing through many advantages, enhances adaptability to complex production environments, and promotes the intelligent development of production management, which has positive and important significance for oilfield exploitation operations.

[0139] In another embodiment of the present application, an electronic device is provided, comprising: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;

[0140] Memory, used to store computer programs;

[0141] The processor is used to implement the automatic mixing control method of oilfield fracturing fluid described in the method embodiment when executing the program stored in the memory.

[0142] The communication bus 1140 mentioned in the above electronic device may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus 1140 may be divided into an address bus, a data bus, a control bus, etc.

[0143] For ease of representation, Figure 4 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0144] The communication interface 1120 is used for communication between the above electronic device and other devices.

[0145] The memory 1130 may include a random access memory (RAM) or a non-volatile memory (non-volatile memory), such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0146] The above-mentioned processor 1110 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0147] Accordingly, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed, can implement each step that can be performed by an electronic device in the above method embodiment.

Claims

1. An automatic mixing control method for oilfield fracturing fluid, characterized in that: include: Obtaining the mixing production data of oilfield fracturing fluid; The mixing production data at least includes: equipment status, environmental factors, and production parameters; the production parameters at least include: temperature, conductivity, initial concentration, and liquid flow rate; The mixed production data is predicted in real time by a concentration prediction model to obtain a concentration prediction value of the oilfield fracturing fluid; the concentration prediction model is trained using actual concentration values ​​of the fracturing fluid obtained under different production conditions; the actual concentration value of the fracturing fluid comes from detection data of multiple pipe network nodes in the production environment; Real-time proportioning information of the oilfield fracturing fluid is generated based on the concentration prediction value, and automatic mixing control of the oilfield fracturing fluid is performed through the real-time proportioning information.

2. The automatic mixing control method for oilfield fracturing fluid according to claim 1, characterized in that: The concentration prediction model at least includes: an extraction layer, a construction layer, a correlation layer, and a prediction layer; The concentration prediction model is used to predict the mixed production data in real time to obtain the concentration prediction value of the oilfield fracturing fluid, including: Extracting compounding production features from the compounding production data through the extraction layer; the types of the compounding production features are dynamically configured based on the production environment and / or production demand; Through the construction layer, the mixed production characteristics are constructed as corresponding production element nodes in the oilfield fracturing fluid association graph network; the production element nodes at least include: equipment status nodes, environmental factor nodes, and production parameter nodes; Through the association layer, the association relationship between the production element nodes is obtained, and a network connection structure between the production element nodes is constructed based on the association relationship; the association relationship at least includes: the spatiotemporal association weight, the ratio concentration association, and the degree of influence on the ratio result between the production element nodes; Through the prediction layer, the ratio result prediction is performed according to the oilfield fracturing fluid association graph network to obtain the concentration prediction value of the oilfield fracturing fluid.

3. The automatic mixing control method for oilfield fracturing fluid according to claim 2, characterized in that: The production element node v is obtained in the association layer i and produce element node v j The spatiotemporal correlation weight W ts The process is expressed as the following expression: in, The square of the standard deviation parameter representing the time correlation, It is used to control the magnitude of the weight change with time delay. t represents the time delay factor corresponding to the current device and production environment. ij Indicates the production element node v in the mixed production data i and produce element node v j The average time delay of the association, α represents the production element node v i and produce element node v j The time correlation weight ratio between Represents the production element node v i and produce element node v j The time correlation weight between them, β represents the production element node v i and produce element node v j The spatial correlation weight ratio between Represents the production element node v in the pipe network system of the actual production equipment i and produce element node v j The spatial correlation weight between ||v i -v j ||2 represents the production element node v in the pipe network system i and produce element node v j The equivalent spatial distance between .

4. The automatic mixing control method for oilfield fracturing fluid according to claim 3, characterized in that: Produce element node v i The set of eigenvalues ​​associated with the ratio concentration is c i , produce element node v j The set of eigenvalues ​​associated with the ratio concentration is c j , the characteristic value set c is obtained in the association layer i and the eigenvalue set c j The combined ratio concentration correlation evaluation value W c The process is expressed as the following expression: Among them, c i t and c j t Respectively represent the eigenvalue set c i and the eigenvalue set c j The eigenvalue in the sample at the tth moment, and Respectively represent the eigenvalue set c i and the eigenvalue set c j The respective sample means, T is the total number of samples.

5. The automatic mixing control method for oilfield fracturing fluid according to claim 4, characterized in that: The step of obtaining the influence degree of the ratio result of the production element node in the association layer includes: A multivariate linear regression analysis model was constructed with the characteristic values ​​of each production element node in the historical mixed production data as the independent variable and the historical fracturing fluid concentration corresponding to each production element node as the target variable. Will produce element node v i Input into the multivariate linear regression analysis model to obtain the production element node v i The regression coefficient vector of each eigenvalue in ; Will produce element node v i The absolute value of the regression coefficient vector of each eigenvalue in is normalized to obtain the production element node v i The corresponding impact assessment value of the matching result.

6. The automatic mixing control method for oilfield fracturing fluid according to claim 5, characterized in that: The step of constructing a network connection structure between the production element nodes based on the association relationship includes: The Gini impurity evaluation method is used to evaluate the production element node v i Comprehensively optimize the spatiotemporal correlation weight, the ratio concentration correlation evaluation value, and the evaluation value of the impact degree on the ratio result; The optimized spatiotemporal correlation weight, the ratio concentration correlation evaluation value, and the evaluation value of the impact degree on the ratio result are mapped to weight parameters to obtain the production element node v i Connection parameters with surrounding nodes; Set and construct the production element node v based on the connection parameters i The associated topology.

7. The automatic mixing control method for oilfield fracturing fluid according to claim 2, characterized in that: The prediction layer predicts the ratio result according to the oilfield fracturing fluid association graph network to obtain the concentration prediction value of the oilfield fracturing fluid, including: Determine the correlation relationship between the production element nodes in the oilfield fracturing fluid correlation graph network, and calculate the corresponding weight matrix; The weight matrix is ​​converted into an adjacency matrix, and the oilfield fracturing fluid association graph network is subjected to information propagation and aggregation operations, and the adjacency matrix is ​​processed to obtain a correlation matrix of the oilfield fracturing fluid association graph network; the correlation matrix at least includes a unit matrix and a degree matrix matched with the adjacency matrix; Optimizing the weight matrix based on the correlation matrix; Using the optimized weight matrix to construct a node feature matrix corresponding to the oilfield fracturing fluid association graph network; The node feature matrix is ​​iteratively updated and fully connected layer calculated to obtain a concentration prediction value of the oilfield fracturing fluid.

8. An automatic mixing control device for oilfield fracturing fluid, characterized in that: The device comprises: An acquisition unit is configured to acquire mixed production data of oilfield fracturing fluid; the mixed production data at least includes: equipment status, environmental factors, production parameters; the production parameters at least include: temperature, conductivity, initial concentration, liquid flow rate; The prediction unit is configured to perform real-time prediction on the mixed production data through a concentration prediction model to obtain a concentration prediction value of the oilfield fracturing fluid; the concentration prediction model is trained using actual concentration values ​​of the fracturing fluid obtained under different production conditions; the actual concentration value of the fracturing fluid comes from detection data of multiple pipe network nodes in the production environment; The execution unit is configured to generate real-time ratio information of the oilfield fracturing fluid based on the concentration prediction value, and execute automatic mixing control of the oilfield fracturing fluid through the real-time ratio information.

9. An electronic device, characterized in that: include: Memory for storing computer software programs; A processor is used to read and execute the computer software program, thereby implementing the automatic mixing control method of oilfield fracturing fluid according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, characterized in that: The storage medium stores a computer software program, and when the computer software program is executed by the processor, the automatic mixing control method for oilfield fracturing fluid according to any one of claims 1 to 7 is implemented.

Images