An adaptive control system for oilfield equipment

Abstract

The invention discloses an adaptive control system for oilfield equipment, which specifically relates to the technical field of equipment regulation and control, and comprises a data acquisition module, which is used to acquire equipment operation data and environmental data, and generate an operation data set and an environmental data set; a data analysis module, which pre-processes the operation data set, extracts key features of the environmental data, and calculates in real time a reservoir characteristic fluctuation index and a crude oil viscosity fluctuation index; a prediction module, which predicts the change trend of the above fluctuation indexes through a time series prediction model; a switching module, which determines whether it is necessary to enable an adaptive control strategy according to the prediction results; and a decision module, which dynamically adjusts equipment operation parameters when the adaptive control strategy is enabled, so as to ensure that the equipment operates efficiently under complex working conditions; the invention perceives working condition changes in advance, avoids equipment efficiency reduction caused by sudden changes in working conditions, and starts the adaptive control strategy when necessary, so as to ensure control accuracy while reducing unnecessary strategy switching, thereby reducing system calculation burden and energy consumption.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment regulation and control, and more specifically, to an oilfield equipment adaptive control system. Background Art

[0002] The operating environment of oilfield equipment is usually complex and changeable. These dynamic changes directly affect the operating efficiency and safety of oilfield equipment. For example, changes in reservoir permeability may lead to untimely adjustment of equipment pump speed, thereby affecting oil production; drastic fluctuations in crude oil viscosity may increase equipment energy consumption and even cause equipment damage. Therefore, traditional fixed control strategies often show insufficient flexibility when dealing with complex working conditions, and it is difficult to maintain efficient operation of equipment under different operating conditions.

[0003] In the existing technology, adaptive control strategies are gradually introduced into the control system of oilfield equipment. Adaptive control dynamically adjusts the control strategy to adapt to complex working conditions by real-time monitoring of equipment operating parameters and environmental data. However, existing adaptive control technologies usually rely on a single data source and lack a comprehensive analysis of reservoir characteristics and crude oil viscosity changes; in addition, the switching of control strategies lacks scientificity, which may lead to a decrease in the stability of equipment operation. Therefore, a more intelligent and flexible adaptive control system for oilfield equipment is needed to better cope with the complex dynamic working conditions of oilfields and improve the efficiency and safety of equipment operation. Summary of the invention

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] An oilfield equipment adaptive control system includes a data acquisition module, a data analysis module, a prediction module, a switching module, and a decision module;

[0006] The data acquisition module is used to collect the operation data and environmental data of the oilfield equipment to obtain the operation data set and the environmental data set;

[0007] The data analysis module is used to pre-process the collected operational data set and extract key characteristic parameters from the environmental data set, and then generate the reservoir characteristic fluctuation index and crude oil viscosity fluctuation index based on real-time calculation;

[0008] The prediction module is used to predict the change trends of the reservoir characteristic fluctuation index and crude oil viscosity fluctuation index respectively through the time series prediction model based on the historical data and the reservoir characteristic fluctuation index and crude oil viscosity fluctuation index calculated in real time;

[0009] The switching module is used to determine whether the adaptive control strategy needs to be enabled based on the prediction results;

[0010] The decision module is used to dynamically adjust the operating parameters of oilfield equipment based on the real-time preprocessed operating data set, the real-time calculated reservoir characteristic fluctuation index and the crude oil viscosity fluctuation index when the adaptive control strategy is enabled, so as to ensure the efficient operation of oilfield equipment under complex working conditions.

[0011] In a preferred embodiment, the environmental data set consists of a reservoir property data set and a crude oil property data set.

[0012] In a preferred embodiment, when the adaptive control strategy is not enabled, the preset fixed control strategy can be used to meet the operation requirements of the oilfield equipment.

[0013] In a preferred embodiment, extracting key characteristic parameters from the environmental data set refers to:

[0014] A plurality of preset reservoir characteristic feature parameters are extracted from the reservoir characteristic data set, and then a reservoir characteristic fluctuation index is generated based on real-time calculation; a plurality of preset crude oil characteristic feature parameters are extracted from the crude oil characteristic data set, and then a crude oil viscosity fluctuation index is generated based on real-time calculation.

[0015] In a preferred embodiment, the logic for obtaining the reservoir property fluctuation index is:

[0016] Obtain each reservoir characteristic parameter to form a time series data set. , and then normalize different physical quantities to eliminate dimensional differences:

[0017] ; represents the reservoir characteristic parameter value corresponding to time point t, , represent the mean and standard deviation of the reservoir characteristic parameters, It represents the reservoir characteristic parameter value corresponding to the time point t after eliminating the dimension difference;

[0018] Calculate the magnitude of change at adjacent time points: ; Indicates the change amplitude corresponding to time point t;

[0019] Calculate the trend deviation using the sliding window method: ; N represents the sliding window size, Indicates the trend deviation corresponding to time point t, represents the average value within the window;

[0020] Introduce time decay weights to enhance the impact of recent data: ; represents the time decay weight corresponding to time point t, represents the preset time attenuation coefficient, and T represents the current time;

[0021] ; represents the preset trend deviation weight coefficient, K represents the end point of the time window, Indicates the fluctuation coefficient corresponding to the characteristic parameters of reservoir characteristics;

[0022] The fluctuation coefficients corresponding to multiple preset reservoir characteristic parameters are summarized to obtain a fluctuation vector, and then the Euclidean distance between the fluctuation vector and the preset reservoir standard vector is calculated to obtain the reservoir characteristic fluctuation index corresponding to the time point t. .

[0023] In a preferred embodiment, the logic for obtaining the crude oil viscosity fluctuation index is:

[0024] Get real-time viscosity data of crude oil , Ambient temperature Environmental pressure , forming time series data set 2 , n represents the time point;

[0025] Calculate the rate of change of viscosity at consecutive time points: ; represents the viscosity change rate corresponding to time point n, Indicates the time interval between adjacent time points;

[0026] Calculate the weighted mean of the viscosity fluctuation amplitude within the sliding window to obtain the viscosity fluctuation amplitude impact value: ; The sliding window size is M, Indicates the influence value of viscosity fluctuation amplitude, represents the mean value of viscosity in the window, represents the time influence weight corresponding to time point j, satisfying the following conditions: ; represents the viscosity change rate corresponding to time point j, represents the viscosity change rate corresponding to the time point p;

[0027] According to the fluctuation range of ambient temperature and pressure, calculate the environmental sensitivity factor:

[0028] ; Indicates the preset temperature influence coefficient, Indicates the preset pressure influence coefficient, and denote the historical average temperature and pressure, respectively. represents the environmental sensitivity factor corresponding to time point n;

[0029] ; , , All are preset adjustment coefficients. Represents the crude oil viscosity fluctuation index corresponding to time point n.

[0030] In a preferred embodiment, the time series prediction model refers to:

[0031] A dynamic fusion deep learning model is adopted, combining convolutional neural network and long short-term memory network, and incorporating attention mechanism to enhance the capture of reservoir characteristic fluctuation trend or crude oil characteristic fluctuation trend, and output the reservoir characteristic fluctuation index and crude oil viscosity fluctuation index at the next time point.

[0032] In a preferred embodiment, judging whether it is necessary to enable the adaptive control strategy based on the prediction results means: based on the predicted reservoir characteristic fluctuation index and crude oil viscosity fluctuation index at the next time point, using fuzzy reasoning, reasoning whether it is necessary to enable the adaptive control strategy, when the obtained reasoning type is one, the adaptive control strategy needs to be enabled, and when the obtained reasoning type is two, the adaptive control strategy does not need to be enabled.

[0033] Technical effects and advantages of the present invention:

[0034] The present invention can monitor the fluctuation of reservoir characteristics and crude oil viscosity in real time, and perceive the change of working conditions in advance through the prediction model, and quickly adjust the equipment operating parameters (such as pump speed, valve opening, etc.). It avoids the decline of equipment efficiency due to sudden changes in working conditions, thereby improving the overall oil production efficiency. The adaptive control strategy is activated when necessary to ensure control accuracy while reducing unnecessary strategy switching, reducing the system calculation burden and energy consumption.

[0035] The reservoir characteristic fluctuation index and crude oil viscosity fluctuation index can accurately quantify the dynamic changes of reservoir and crude oil status, and identify the risks that may cause equipment overload or failure in advance. The system adjusts the operation strategy in time under complex working conditions to avoid equipment damage due to severe fluctuations and extend the service life of the equipment. Fuzzy reasoning is introduced, combined with historical data and real-time fluctuation characteristics, to achieve scientific judgment on the switching of control strategies and reduce the operational risks caused by human intervention.

[0036] The present invention not only collects equipment operation data, but also analyzes the dynamic changes of reservoir characteristics and crude oil viscosity, providing a more comprehensive working condition assessment. It can adapt to the reservoir conditions and crude oil characteristics of different oil fields without additional adjustment of hardware equipment. The system fully considers the impact of fluctuations in ambient temperature and pressure on equipment operation, and enhances the adaptability of the equipment in extreme environments by calculating environmental sensitivity factors. The system intelligently determines whether it is necessary to enable the adaptive control strategy based on the results of fuzzy reasoning. When the working conditions are stable, the fixed strategy is maintained, and when the working conditions are complex, the adaptive control strategy is switched. The system instability caused by frequent switching of strategies is avoided, while the flexibility of the equipment under special working conditions is ensured. By using fixed strategies under stable working conditions, the calculation and adjustment costs of adaptive control are reduced, and the long-term operating cost of the system is optimized. Convolutional neural networks, long short-term memory networks and attention mechanisms are introduced to accurately capture the dynamic trends of reservoir characteristics and crude oil viscosity. Combined with fuzzy reasoning, an intelligent closed loop from prediction to decision-making is realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;

[0038] Figure 1 The schematic diagram of an adaptive control system for oilfield equipment in the present invention. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] Reference Figure 1 The following embodiments are obtained:

[0041] Example 1

[0042] An oilfield equipment adaptive control system includes a data acquisition module, a data analysis module, a prediction module, a switching module, and a decision module;

[0043] The data acquisition module is used to collect the operating data of oilfield equipment and environmental data to obtain the operating data set and environmental data set; through the data acquisition module, the operating status information (such as speed, flow, etc.) and environmental condition data (such as reservoir characteristics, crude oil viscosity) of the oilfield equipment are obtained in real time to generate the operating data set and environmental data set. It provides basic data support for subsequent data analysis and decision-making, ensuring that the system has comprehensive and accurate input information. Data acquisition is the starting point of the entire system, and its quality directly affects the reliability and accuracy of subsequent modules.

[0044] The data analysis module is used to preprocess the collected operating data set and extract key characteristic parameters from the environmental data set. Then, based on real-time calculation, the reservoir characteristic fluctuation index and crude oil viscosity fluctuation index are generated respectively. The collected operating data set is preprocessed (such as denoising and normalization) to clean invalid or abnormal data and improve the availability and accuracy of the data. Key characteristic parameters (such as reservoir pressure changes and crude oil viscosity fluctuations) are extracted from the environmental data set to lay the foundation for the generation of the fluctuation index. The reservoir characteristic fluctuation index and crude oil viscosity fluctuation index are generated by real-time calculation. These two core indexes can quantitatively reflect the dynamic changes of complex working conditions at the oil field site and provide high-value information for prediction and decision-making.

[0045] The prediction module is used to predict the change trends of the reservoir characteristic fluctuation index and crude oil viscosity fluctuation index respectively based on historical data and real-time calculation through the time series prediction model; based on the reservoir characteristic fluctuation index and crude oil viscosity fluctuation index obtained by historical data and real-time calculation, the time series prediction model is used to predict their future change trends. It can identify the possible drastic fluctuations in reservoir characteristics and crude oil viscosity in advance, help the system formulate response strategies in advance, and avoid equipment operating efficiency reduction or damage caused by sudden changes in operating conditions. It can improve the system's foresight and dynamic adaptability, and provide a scientific basis for the judgment of switching modules.

[0046] The switching module is used to determine whether the adaptive control strategy needs to be enabled based on the prediction results; based on the results of the prediction module, it determines whether the working conditions exceed the coping capacity of the fixed control strategy, and switches to the adaptive control strategy when necessary. When the working condition fluctuates slightly, the fixed control strategy is maintained to reduce the computational burden and resource consumption; when the working condition fluctuates greatly, the adaptive control strategy is enabled to ensure that the system adapts to complex changes. The switching module is the central judgment unit of the system, which can dynamically balance the stability and flexibility of the system.

[0047] The decision module is used to dynamically adjust the operating parameters of oilfield equipment based on the real-time preprocessed operating data set, the reservoir characteristic fluctuation index and the crude oil viscosity fluctuation index calculated in real time when the adaptive control strategy is enabled, so as to ensure the efficient operation of oilfield equipment under complex working conditions. When the adaptive control strategy is enabled, the operating parameters of oilfield equipment (such as pump speed, valve opening, etc.) are dynamically adjusted according to real-time data and calculation results. Ensure that the equipment operates in an efficient and safe state under complex working conditions, maximize production efficiency, and reduce the risk of equipment failure. The decision module is the execution unit of the system, and its function directly determines the actual performance of the entire system under complex working conditions.

[0048] Unlike traditional fixed control strategies, adaptive control has higher flexibility and intelligence. It can optimize the equipment's operating strategy in real time without human intervention according to the complex and changeable working conditions at the oil field site (such as changes in reservoir characteristics or fluctuations in crude oil viscosity). This not only improves the production efficiency of the equipment, but also reduces the risk of failure caused by overload or parameter mismatch, while saving energy and reducing operating costs, providing technical support for the efficient, safe and sustainable development of oil field production.

[0049] When the impact of fluctuations in working conditions is small, the operating state of the equipment is basically within the preset parameter range, and the fixed control strategy can meet production needs. This method avoids the additional computational overhead of enabling adaptive control algorithms and can maintain the simplicity and stability of the system. Especially when the operating conditions are relatively stable, the fixed strategy can provide reliable control effects with lower complexity.

[0050] When the fluctuation of working conditions has a great impact, the fixed control strategy may not meet the production requirements, such as reducing production efficiency or increasing the risk of equipment damage. At this time, enabling the adaptive strategy can dynamically adjust the parameters to adapt to the new working conditions and ensure the stable and efficient operation of the equipment. This mechanism of selecting control strategies based on the degree of fluctuation is a compromise between complexity and adaptability, which saves resources and can cope with complex changes.

[0051] Adaptive control algorithms usually require high computing power. If the adaptive strategy is always enabled, even if the fluctuation is small, it will increase unnecessary computing and energy consumption, resulting in a waste of resources. The preset fixed control strategy is based on stable preset parameters and is suitable for stable operation scenarios after sufficient verification. Enabling adaptive control may introduce a dynamic adjustment process, and excessive or unnecessary adjustments may cause short-term instability of the system. Using a fixed strategy in the case of small fluctuations can reduce the complexity of maintenance and troubleshooting and improve reliability. The operating environment of some oilfield equipment may be relatively stable most of the time, with occasional large fluctuations. If adaptive control is always enabled, it may not be cost-effective.

[0052] The environmental data set consists of a reservoir characteristic data set and a crude oil characteristic data set. When the adaptive control strategy is not enabled, the preset fixed control strategy can meet the oilfield equipment operation requirements. Extracting key characteristic parameters from the environmental data set means: extracting multiple preset reservoir characteristic characteristic parameters from the reservoir characteristic data set, and then generating a reservoir characteristic fluctuation index based on real-time calculation; extracting multiple preset crude oil characteristic characteristic parameters from the crude oil characteristic data set, and then generating a crude oil viscosity fluctuation index based on real-time calculation.

[0053] The division of environmental data sets clarifies the key external factors of oilfield operation, including reservoir characteristics (such as permeability, pressure, porosity) and crude oil characteristics (such as viscosity affected by environmental factors). These two parts of data jointly affect the operating status and operating condition changes of oilfield equipment. Through this classification, parameters can be extracted and analyzed for different characteristic data respectively, and the impact of environmental changes can be evaluated and responded to more specifically, improving the accuracy and reliability of system analysis. When the adaptive control strategy is not enabled, the preset fixed control strategy can meet the operating requirements of oilfield equipment. When the environmental changes are small and the operating conditions are stable, the preset fixed control strategy can meet the equipment operation requirements without dynamically adjusting the equipment parameters, avoiding the computing resource consumption of the adaptive control strategy. This design not only reduces the complexity and computing overhead of the system operation, but also enhances the stability of the system, especially in the case of long-term stable operation. Through this mode, the system realizes "on-demand regulation" and achieves the optimal utilization of resources while meeting the operating requirements.

[0054] Extract multiple preset reservoir characteristic parameters from the reservoir characteristic data set and generate a reservoir characteristic fluctuation index. Reservoir characteristic parameters (such as permeability, pressure, and porosity) are key indicators of reservoir dynamic changes. Extracting these parameters can effectively capture the fluctuation characteristics and change trends of the reservoir. The generated reservoir characteristic fluctuation index is a comprehensive quantification of these parameters, which can reflect the impact of the reservoir on the operation of oilfield equipment. For example, when the reservoir pressure fluctuates violently, the fluctuation index increases, which can timely identify and predict the potential risks of reservoir changes to equipment. This step-by-step extraction and comprehensive calculation method ensures the scientific nature of the analysis process and the credibility of the results, providing an accurate basis for subsequent decision-making.

[0055] Extract multiple preset crude oil characteristic parameters from the crude oil characteristic data set and generate the crude oil viscosity fluctuation index. Crude oil characteristics (such as viscosity affected by environmental factors) directly affect the fluidity of crude oil and the energy consumption level of equipment. Extracting these characteristic parameters can fully capture the changes in crude oil state, especially the changing trend of crude oil viscosity. The crude oil viscosity fluctuation index can reflect the changes in crude oil fluidity and its potential impact on equipment operation through real-time calculation and comprehensive quantification of these parameters. For example, when the viscosity of crude oil suddenly increases, the fluctuation index will rise, which prompts the equipment to adjust the pump speed or heating power to ensure operating efficiency. This design enables the system to perceive the drastic changes in the crude oil state in advance, so as to actively adjust the equipment operating parameters and reduce the impact of sudden conditions on system operation.

[0056] By extracting preset characteristic parameters from the reservoir characteristic data set and the crude oil characteristic data set, the interference of irrelevant or secondary data can be reduced, and the analysis efficiency and accuracy can be improved. After extracting the characteristic parameters, the reservoir characteristic fluctuation index and the crude oil viscosity fluctuation index are generated based on real-time calculation, which can dynamically reflect the changes in environmental data and ensure the adaptability of the system under complex working conditions. The calculation result of the fluctuation index is the dividing line between the fixed control strategy and the adaptive control strategy. By quantifying the fluctuation of the reservoir and crude oil, the system can intelligently determine whether the adaptive control strategy needs to be enabled to avoid unnecessary switching or waste of resources.

[0057] The logic for obtaining the reservoir characteristic fluctuation index is:

[0058] Obtain each reservoir characteristic parameter to form a time series data set. , and then normalize different physical quantities to eliminate dimensional differences:

[0059] ; represents the reservoir characteristic parameter value corresponding to time point t, , represent the mean and standard deviation of the reservoir characteristic parameters, It represents the reservoir characteristic parameter value corresponding to the time point t after eliminating the dimension difference; the normalized data have the same scale in statistical distribution, and the normalization step ensures that the subsequent fluctuation calculation is not affected by the numerical range.

[0060] Calculate the magnitude of change at adjacent time points: ; It indicates the change amplitude corresponding to time point t; it calculates the normalized parameter difference between adjacent time points, and takes the absolute value to reflect the magnitude of the fluctuation, which is used to quantify the change amplitude of reservoir parameters in a short period of time and directly reflects the dynamic changes of reservoir characteristic parameters in a short period of time.

[0061] Calculate the trend deviation using the sliding window method: ; N represents the sliding window size, Indicates the trend deviation corresponding to time point t, It represents the average value within the window; trend deviation is an important feature for capturing medium-term fluctuations, reflecting the dynamic deviation between reservoir characteristic parameters and trends.

[0062] Introduce time decay weights to enhance the impact of recent data: ; represents the time decay weight corresponding to time point t, Represents the preset time decay coefficient, T represents the current time; the purpose is to make the data closer to the current time have a higher weight, and λ controls the decay speed of the weight, so that more attention can be paid to the latest data and the impact of historical data on the index calculation can be reduced.

[0063] ; represents the preset trend deviation weight coefficient, K represents the end point of the time window, It represents the fluctuation coefficient corresponding to the characteristic parameters of reservoir characteristics; it combines short-term fluctuations and medium-term trend deviations, and dynamically adjusts through time weights to reflect the comprehensive dynamics of reservoir characteristic fluctuations.

[0064] For multiple preset reservoir characteristic parameters (such as permeability, pressure, porosity), their fluctuation coefficients are calculated respectively, and the fluctuation coefficients corresponding to multiple preset reservoir characteristic parameters are summarized to obtain a fluctuation vector, and then the Euclidean distance between the fluctuation vector and the preset reservoir standard vector is calculated to obtain the reservoir characteristic fluctuation index corresponding to the time point t , the larger the distance, the larger the reservoir characteristic fluctuation index, indicating that the current reservoir characteristic deviates more seriously from the standard state. The fluctuation vector can fully reflect the dynamic changes of reservoir characteristics at time point t. The fluctuation coefficients of each characteristic parameter are integrated into one vector, avoiding the influence of a single parameter on the overall analysis and providing more comprehensive reservoir state information. The standard vector represents the reservoir characteristic state under ideal working conditions and is the benchmark for measuring the degree of deviation of the current reservoir state. This standard can be set based on the statistical results of historical data or expert experience to ensure that the results are highly correlated with actual working conditions.

[0065] The logic for obtaining the crude oil viscosity fluctuation index is:

[0066] Get real-time viscosity data of crude oil , Ambient temperature Environmental pressure , forming time series data set 2 , n represents the time point;

[0067] Calculate the rate of change of viscosity at consecutive time points: ; represents the viscosity change rate corresponding to time point n, Indicates the time interval between adjacent time points; It is used to quantify the dynamic change of crude oil viscosity in a short period of time. The viscosity change rate is a key indicator reflecting the change of crude oil fluidity and has direct significance for the pump speed, energy consumption and fluidity regulation of oilfield equipment. Violent fluctuations indicate that the crude oil viscosity state is unstable, which may affect the operating efficiency of the equipment.

[0068] Calculate the weighted mean of the viscosity fluctuation amplitude within the sliding window to obtain the viscosity fluctuation amplitude impact value: ; The sliding window size is M, Indicates the viscosity fluctuation amplitude impact value, which is used to calculate the impact of viscosity fluctuation amplitude on the system. Indicates the mean value and fluctuation range of viscosity within the window Indicates the difference between the current viscosity value and the mean value in the window, reflecting the local fluctuation characteristics of the viscosity. represents the viscosity of crude oil corresponding to time point j. The sliding window mechanism M controls the balance between short-term and long-term trends, making It is more suitable for describing the change characteristics of viscosity in local time. represents the time influence weight corresponding to time point j, satisfying the following conditions: ; represents the viscosity change rate corresponding to time point j, represents the viscosity change rate corresponding to the time point p;

[0069] According to the fluctuation range of ambient temperature and pressure, calculate the environmental sensitivity factor:

[0070] ; Indicates the preset temperature influence coefficient, Indicates the preset pressure influence coefficient, allowing adjustment of the influence of different environmental parameters. and denote the historical average temperature and pressure, respectively. Represents the environmental sensitivity factor corresponding to time point n; the environmental sensitivity factor is calculated by the fluctuation amplitude of ambient temperature and pressure, directly reflecting the sensitivity of environmental changes to the viscosity state of crude oil, and is the key input for further calculation of the comprehensive volatility index. Temperature and pressure are the main external factors affecting the change of crude oil viscosity. By normalizing the fluctuation amplitude of these two parameters, their impact on the system can be quantified.

[0071] ; , , All are preset adjustment coefficients. Indicates the crude oil viscosity fluctuation index corresponding to time point n. Adjustment coefficient of fluctuation amplitude , used to emphasize the effect of viscosity fluctuations. Rate adjustment coefficient , used to highlight the dynamic effect of the viscosity change rate. Adjustment coefficient for environmental sensitivity , which is used to comprehensively consider the potential impact of the external environment on the system. The crude oil viscosity fluctuation index realizes the comprehensive quantification of the crude oil viscosity state and can accurately reflect the dynamic fluctuation of crude oil characteristics.

[0072] It should be noted that the time points in the calculation process of crude oil viscosity fluctuation index and reservoir characteristic fluctuation index are consistent, which ensures the reasoning quality after the subsequent time series prediction model prediction.

[0073] Time series forecasting models refer to:

[0074] A dynamic fusion deep learning model is adopted, combining convolutional neural network (CNN) and long short-term memory network (LSTM), and incorporating attention mechanism to enhance the capture of reservoir property fluctuation trend or crude oil property fluctuation trend, and output the reservoir property fluctuation index and crude oil viscosity fluctuation index at the next time point.

[0075] Specifically, the historical data of the reservoir characteristic fluctuation index or the crude oil viscosity fluctuation index are obtained to obtain the time series data set 3 corresponding to the reservoir characteristic fluctuation index or the time series data set 4 corresponding to the crude oil viscosity fluctuation index. Take this as an example to illustrate: yes After normalization.

[0076] Feature extraction (CNN model): Input time series dataset three To the CNN module, extract local fluctuation features: ; k is the convolution kernel size, which indicates the number of time points included in each calculation when sliding on the time series. s is the step size, which indicates the number of interval steps of the convolution kernel sliding on the time series. It is a local fluctuation feature. After the convolution operation, the local pattern or short-term dynamic change feature in the time series is extracted as the input of LSTM to further capture the temporal dependency of the time series. ReLU is an activation function, which is often used in deep learning models. Its function is to convert all negative values ​​to zero and retain positive values, thereby increasing nonlinear capabilities. Conv1D represents a one-dimensional convolution operation, which is used to process one-dimensional data sequences such as time series data. The convolution operation extracts features through a sliding window.

[0077] Time Series Modeling (LSTM Model): Extracted Local Features Enter the LSTM model to capture long-term and short-term dependencies: ;

[0078] Attention mechanism: Introduce the attention mechanism to calculate the weight of each time step, that is, the attention weight at time t : ; Here t represents the historical time point index in the time series used to calculate the attention weight, ranging from 1 to G, where G is the length of the time window and represents the number of historical data points used for model calculation. ; , Represents trainable parameters, representing the weight and bias of the linear transformation, which are used to learn the attention weights, where is a trainable weight matrix, representing the hidden state The linear mapping of is a trainable bias vector used to adjust the calculation of the attention score.

[0079] The weighted summation gives the context vector C: ; Represents the attention score at time t, indicating the relevance of the feature at the current time point to the predicted target. Represents the hidden state at time t, which represents the features of the current time point captured by the LSTM model and contains both short-term and long-term dependency information. is the attention weight at time t, indicating the importance of the current time point to the overall context. The context vector C represents the weighted comprehensive information of the time series features. The context vector combines the information at different time points through weighted summation, and the weights are dynamically allocated by the attention mechanism.

[0080] Output layer: The context vector C is input into the fully connected layer and the predicted value at the next time point is output: ; is the predicted value output by the model, which indicates the predicted result of the reservoir characteristic fluctuation index at the next time point. FC is a fully connected layer, which is used to map the context vector C to a specific predicted value. When optimizing the loss function, the weighted mean square error is used as the loss function to improve the sensitivity to abnormal points. I will not go into details here. The model is trained with historical data to predict the predicted value at the next time point, that is, the reservoir characteristic fluctuation index. The prediction method of the crude oil viscosity fluctuation index is the same as the above principle. Combining the local feature extraction capability of CNN, the temporal dependency modeling capability of LSTM, and the dynamic weight allocation capability of the attention mechanism, the accuracy and adaptability of the reservoir characteristic fluctuation index prediction are effectively improved.

[0081] Judging whether it is necessary to enable the adaptive control strategy based on the prediction results means: based on the predicted reservoir characteristic fluctuation index and crude oil viscosity fluctuation index at the next time point, using fuzzy reasoning, reasoning whether it is necessary to enable the adaptive control strategy, when the obtained reasoning type is one, the adaptive control strategy needs to be enabled, and when the obtained reasoning type is two, the adaptive control strategy does not need to be enabled.

[0082] During the operation of oilfield equipment, dynamic changes in the environment will lead to instability in reservoir characteristics and crude oil viscosity. In order to ensure the efficient operation of the equipment, it is necessary to determine whether a more flexible adaptive control strategy needs to be enabled based on the predicted reservoir characteristic fluctuation index and crude oil viscosity fluctuation index to cope with possible complex working conditions.

[0083] The system calculates the reservoir characteristic fluctuation index and crude oil viscosity fluctuation index at the next time point through the prediction model. These two indexes reflect the dynamic change degree of the reservoir and crude oil state. Fuzzy reasoning is an intelligent decision-making method based on rules and fuzzy logic. The system takes the two predicted fluctuation indices as input and makes logical judgments according to the pre-set reasoning rules.

[0084] Convert the input precise values ​​(such as reservoir property volatility index and crude oil viscosity volatility index) into fuzzy linguistic variables. Linguistic variables are a collection of fuzzy sets, such as "small volatility", "medium volatility", and "large volatility". Each input value is mapped to one or more fuzzy sets according to the membership function to obtain the corresponding membership degree.

[0085] The fuzzy rule base defines rules according to specific application scenarios, such as: if the reservoir characteristic fluctuation is "large fluctuation" and the crude oil viscosity fluctuation is "large fluctuation", then the adaptive control strategy needs to be enabled. If the reservoir characteristic fluctuation is "small fluctuation" and the crude oil viscosity fluctuation is "medium fluctuation", then the adaptive control strategy does not need to be enabled. The fuzzy rule base can be generated based on expert experience, historical data or self-learning.

[0086] Use fuzzy logic operations to match input variables with fuzzy rules. According to the rules that meet the conditions in the rule base, combined with the membership of the input variables, the fuzzy set membership of the output variable is calculated. The main reasoning methods include: maximum and minimum reasoning method (Min-Max method), integral reasoning method, centroid method, etc.

[0087] Convert the fuzzy output variables obtained by reasoning (such as "need to adjust" and "no need to adjust") into specific numerical results. Common defuzzification methods include: Center of gravity method: Calculate the weighted average of the output value according to the membership distribution of the fuzzy set. Maximum membership method: Select the fuzzy set with the highest membership as the output.

[0088] Define the fuzzification range and language description of the input variables and output variables. For example, "small fluctuation" can be defined as an index in the range of 0 to 0.3, and "large fluctuation" can be defined as an index in the range of 0.6 to 1. Each fuzzy set has a corresponding membership function, which is used to quantify the degree to which a certain input value belongs to the fuzzy set. The membership function is usually a trigonometric function, a trapezoidal function, or a Gaussian function. The rule base stores the knowledge of fuzzy reasoning, which is usually a summary of expert experience or data analysis. The fuzzy inference engine realizes the mapping from input to output and performs reasoning based on the rule base. The reasoning results are divided into two types:

[0089] Type 1: The operating conditions are complex and fluctuate significantly, and the adaptive control strategy needs to be enabled. At this time, the equipment will dynamically adjust the operating parameters to adapt to complex changes and improve operating efficiency and safety.

[0090] Type 2: The operating conditions are stable and the fluctuations are small, so there is no need to enable the adaptive control strategy. The device will continue to use the fixed control strategy to reduce the computing burden and energy consumption.

[0091] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0092] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0093] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0094] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0095] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An oilfield equipment adaptive control system, characterized in that: It includes data acquisition module, data analysis module, prediction module, switching module and decision-making module; The data acquisition module is used to collect the operation data and environmental data of the oilfield equipment to obtain the operation data set and the environmental data set; The data analysis module is used to pre-process the collected operational data set and extract key characteristic parameters in the environmental data set, and then generate the reservoir characteristic fluctuation index and crude oil viscosity fluctuation index based on real-time calculation; The prediction module is used to predict the change trends of the reservoir characteristic fluctuation index and crude oil viscosity fluctuation index respectively through the time series prediction model based on the historical data and the reservoir characteristic fluctuation index and crude oil viscosity fluctuation index calculated in real time; The switching module is used to determine whether the adaptive control strategy needs to be enabled based on the prediction results; The decision module is used to dynamically adjust the operating parameters of oilfield equipment based on the real-time preprocessed operating data set, the real-time calculated reservoir characteristic fluctuation index and the crude oil viscosity fluctuation index when the adaptive control strategy is enabled, so as to ensure the efficient operation of oilfield equipment under complex working conditions.

2. An oilfield equipment adaptive control system according to claim 1, characterized in that: The environmental data set consists of a reservoir property data set and a crude oil property data set.

3. An oilfield equipment adaptive control system according to claim 2, characterized in that: When the adaptive control strategy is not enabled, the preset fixed control strategy can meet the operation requirements of oilfield equipment.

4. An oilfield equipment adaptive control system according to claim 3, characterized in that: The key feature parameters extracted from the environmental data set refer to: Extracting a plurality of preset reservoir characteristic feature parameters from the reservoir characteristic data set, and then generating a reservoir characteristic fluctuation index based on real-time calculation; A plurality of preset crude oil characteristic parameters are extracted from the crude oil characteristic data set, and then a crude oil viscosity fluctuation index is generated based on real-time calculation.

5. An oilfield equipment adaptive control system according to claim 4, characterized in that: The logic for obtaining the reservoir characteristic fluctuation index is: Obtain each reservoir characteristic parameter to form a time series data set. , and then normalize different physical quantities to eliminate dimensional differences: ; represents the reservoir characteristic parameter value corresponding to time point t, , represent the mean and standard deviation of the reservoir characteristic parameters, It represents the reservoir characteristic parameter value corresponding to the time point t after eliminating the dimension difference; Calculate the magnitude of change at adjacent time points: ; Indicates the change amplitude corresponding to time point t; Calculate the trend deviation using the sliding window method: ; N represents the sliding window size, Indicates the trend deviation corresponding to time point t, represents the average value within the window; Introduce time decay weights to enhance the impact of recent data: ; represents the time decay weight corresponding to time point t, represents the preset time attenuation coefficient, and T represents the current time; ; represents the preset trend deviation weight coefficient, K represents the end point of the time window, Indicates the fluctuation coefficient corresponding to the characteristic parameters of reservoir characteristics; The fluctuation coefficients corresponding to multiple preset reservoir characteristic parameters are summarized to obtain a fluctuation vector, and then the Euclidean distance between the fluctuation vector and the preset reservoir standard vector is calculated to obtain the reservoir characteristic fluctuation index corresponding to the time point t. .

6. An oilfield equipment adaptive control system according to claim 5, characterized in that: The logic for obtaining the crude oil viscosity fluctuation index is: Get real-time viscosity data of crude oil , Ambient temperature , Environmental Pressure , forming time series data set 2 , n represents the time point; Calculate the rate of change of viscosity at consecutive time points: ; represents the viscosity change rate corresponding to time point n, Indicates the time interval between adjacent time points; Calculate the weighted mean of the viscosity fluctuation amplitude within the sliding window to obtain the viscosity fluctuation amplitude impact value: ; The sliding window size is M, Indicates the influence value of viscosity fluctuation amplitude, represents the mean value of viscosity in the window, represents the crude oil viscosity corresponding to time point j, represents the time influence weight corresponding to time point j, satisfying the following conditions: ; represents the viscosity change rate corresponding to time point j, represents the viscosity change rate corresponding to the time point p; According to the fluctuation range of ambient temperature and pressure, calculate the environmental sensitivity factor: ; Indicates the preset temperature influence coefficient, Indicates the preset pressure influence coefficient, and denote the historical average temperature and pressure, respectively. represents the environmental sensitivity factor corresponding to time point n; ; , , All are preset adjustment coefficients. Represents the crude oil viscosity fluctuation index corresponding to time point n.

7. An oilfield equipment adaptive control system according to claim 6, characterized in that: Time series forecasting models refer to: A dynamic fusion deep learning model is adopted, combining convolutional neural network and long short-term memory network, and incorporating attention mechanism to enhance the capture of reservoir characteristic fluctuation trend or crude oil characteristic fluctuation trend, and output the reservoir characteristic fluctuation index and crude oil viscosity fluctuation index at the next time point.

8. An oilfield equipment adaptive control system according to claim 7, characterized in that: Judging whether it is necessary to enable the adaptive control strategy based on the prediction results means: based on the predicted reservoir characteristic fluctuation index and crude oil viscosity fluctuation index at the next time point, using fuzzy reasoning, reasoning whether it is necessary to enable the adaptive control strategy, when the obtained reasoning type is one, the adaptive control strategy needs to be enabled, and when the obtained reasoning type is two, the adaptive control strategy does not need to be enabled.

Images