Oil reservoir digital twin modeling system and method based on trusted computing

Abstract

The application discloses a kind of oil reservoir digital intelligence twin modeling systems based on trusted computing, including the physical entity layer for real-time acquisition of oil reservoir field data in turn data transmission, responsible for the data processing layer of the data collected, for building the digital twin model of oil reservoir and carrying out model prediction and optimization Intelligent twin layer, realize the data visualization of digital application layer;It also includes to realize the security of data throughout the trusted computing layer;The trusted computing layer is connected with physical entity layer, data processing layer, intelligent twin layer and digital application layer.The system can enhance the security and traceability of data, ensure the security and integrity of data in the process of acquisition, transmission and processing, and also can realize more efficient real-time monitoring and decision support.

Description

Technical Field

[0001] The present invention relates to the field of trusted computing and reservoir management, and more specifically, to a digital twin modeling system and method for oil reservoirs based on trusted computing. Background Art

[0002] Traditional reservoir management practices rely primarily on geological and engineering data to monitor and predict reservoir behavior. Conventional methods, including field sampling, laboratory analysis, and physical monitoring, have historically been industry standards, but they have significant limitations. Specifically, these traditional technologies, due to their reliance on manual labor, result in lengthy and costly data collection cycles. More critically, these methods often lack real-time data updates, making them difficult to adapt to rapidly changing oilfield environments and the need for immediate decision-making. Furthermore, traditional methods face challenges in data accuracy and reliability, particularly in extreme or complex geological conditions.

[0003] With the rapid development of information technology, particularly the application of technologies such as the Internet of Things, big data, artificial intelligence, and machine learning, reservoir management methods and technologies are undergoing a gradual evolution. The introduction of these technologies not only improves the real-time and comprehensive nature of data collection but also significantly enhances the depth and breadth of data analysis. However, effectively integrating and applying these new technologies to reservoir management to achieve efficient data processing and intelligent decision-making remains a major challenge in current research and application.

[0004] Digital twins, an emerging technology concept, create a virtual digital replica of an entity, enabling simulation, prediction, and optimization of its behavior in a virtual environment. In the field of reservoir management, establishing a digital twin model of a reservoir enables real-time monitoring and high-precision analysis of multidimensional reservoir data, significantly improving operational efficiency and effectiveness.

[0005] ① The patent "A Method and Apparatus for Building a Digital Twin of an Oil Reservoir" (patent number CN202310488952.1) filed by China University of Petroleum (East China) is a reservoir-specific digital twin construction method. It includes the following steps: Step 1: Constructing the reservoir physical layer, acquiring reservoir production data through sensors and performing a hierarchical division of physical entities; Step 2: At the reservoir model layer, using 3D Max and Unity to build a geometric model, integrating physical, behavioral, and rule-based models; Step 3: At the reservoir data layer, using a distributed database and a time-series database to manage data based on data update rates; Step 4: At the reservoir connection layer, leveraging the OPC UA standard to enable real-time data exchange between hardware and software; Step 5: At the service application layer, using Unity and external tools such as Matlab for data analysis and real-time prediction to optimize production plans. A drawback of this method is that, while it mentions a combination of "soft sensing" and "hard sensing" to acquire real-time reservoir data, the description of data security and reliability is rather general, focusing primarily on data acquisition and virtual modeling without further detailing data security management and protection measures.

[0006] ② The patent application for the "Trusted Device Management and Control Platform Based on Trusted Computing" (Patent No. 202210830342.0) filed by the 716th Research Institute of China Shipbuilding Industry Corporation (CSIC) is a device management system based on trusted computing technology. It includes the following steps: Step 1: Build a trusted device access layer, connecting trusted devices such as trusted switches, trusted engineering stations, trusted hosts, and trusted servers to provide data support for the platform. Step 2: Set up a basic software layer to provide the platform with the required operating environment, ensure the normal operation of the software layer, and support upper-layer applications. Step 3: Develop a business service layer to implement trusted management, monitoring, auditing, and alerting for trusted devices and resources. Step 4: Develop an application layer, including three modules: a trusted management center, a trusted agent, and a trusted authentication center, to provide the platform's trusted service functions. Step 5: Design a presentation layer to visualize the status of trusted devices and assets, enabling users to intuitively and efficiently manage security in a unified manner. However, this approach may have limitations in processing real-time data and dynamically adjusting system behavior, particularly in its ability to respond to emergencies in highly volatile industrial environments.

[0007] ③ Qiangli IoT Portfolio 2016 Co., Ltd.'s patent, "Intelligent Vibration Digital Twin System and Method for Industrial Environments" (Application Number: CN202080094528.3), describes a platform for updating one or more attributes of one or more digital twins, comprising: Step 1: Receiving a request for one or more digital twins; Step 2: Retrieving the one or more digital twins required to satisfy the request from a digital twin data store; Step 3: Retrieving one or more dynamic models corresponding to the one or more attributes described in the one or more digital twins indicated by the request; Step 4: Selecting a data source from a set of available data sources based on one or more inputs of the one or more dynamic models; Step 5: Obtaining data from the selected data source; Step 6: Determining one or more outputs of the one or more dynamic models using the retrieved data as one or more inputs; Step 7: Updating the one or more attributes of the one or more digital twins based on the one or more outputs of the one or more dynamic models. A disadvantage of this method is that it may be highly dependent on the quality and reliability of external data sources. If the data source is inaccurate or outdated, the updated digital twin may not reflect the true or up-to-date situation.

[0008] Despite this, the application of existing digital twin technology in reservoir management still faces several challenges, such as data security and reliability, the complexity of data processing, and the technical requirements for real-time data processing. Furthermore, ensuring system reliability and data consistency, as well as processing and analyzing large, heterogeneous datasets, are key issues that current technology needs to address. Summary of the Invention

[0009] The present invention aims to address the deficiencies of existing reservoir management technologies in terms of data real-time performance, processing accuracy, and security, and proposes a digital twin modeling system and method for reservoirs based on trusted computing.

[0010] The technical solution of the present invention is:

[0011] A digital twin modeling system for oil reservoirs based on trusted computing includes a physical entity layer for sequentially transmitting data for real-time collection of reservoir field data, a data processing layer responsible for processing the collected data, an intelligent twin layer for constructing a digital twin model of the oil reservoir and performing model prediction and optimization, and a digital intelligence application layer for realizing data visualization; it also includes a trusted computing layer for achieving full-process data security; the trusted computing layer is connected to the physical entity layer, data processing layer, intelligent twin layer, and digital intelligence application layer.

[0012] Preferably, the physical entity layer includes a multi-parameter sensor, an acoustic sensor, an optical fiber sensor and a data acquisition system for real-time acquisition of reservoir environmental data. The data acquisition system is connected to the multi-parameter sensor, the acoustic sensor and the optical fiber sensor. The multi-parameter sensor can at least measure the temperature, pressure, production and chemical composition concentration of the reservoir. The unit of temperature is ℃, the unit of pressure is MPa, and the unit of production is m 3 / d. The unit of chemical component concentration is ppm or mol / L.

[0013] Preferably, the data processing layer includes:

[0014] a decision tree module for classifying collected reservoir environmental data;

[0015] InfluxDB, a time series database for storing reservoir environmental data classified by a decision tree module;

[0016] a Kalman filter module for optimizing the data fusion process;

[0017] A data compression module that uses the Brotli data compression algorithm to reduce the bandwidth and space occupied by data during storage and transmission;

[0018] A data transmission module that uses the MQTT (Message Queuing Telemetry Transport) protocol to ensure efficient and reliable data transmission;

[0019] The decision tree module is connected to the data acquisition system, and the data transmission module is connected to the intelligent twin layer.

[0020] Preferably, the intelligent twin layer constructs a reservoir dynamic behavior model through a recurrent neural network (RNN), uses the synthetic minority oversampling technique (SMOTE) to solve the problem of imbalanced training data in the reservoir dynamic behavior model, and uses the particle swarm optimization (PSO) algorithm to optimize the parameters of the reservoir dynamic behavior model to achieve adaptive learning of the model. The model is regularly verified and updated with new data sets to ensure the accuracy and stability of the model.

[0021] The recurrent neural network RNN ​​constructs a reservoir dynamic behavior model by simulating and predicting the fluid dynamics and multiphase flow characteristics in the reservoir and capturing the evolution of the reservoir state over time; the fluid dynamics and multiphase flow characteristics include oil-water interface changes, pressure and production, with units of m, MPa, m 3 / d;

[0022] The synthetic minority oversampling technique SMOTE is Generate new data sets to address the insufficiency or imbalance of the training data sets and improve the model's prediction and generalization capabilities; For a new dataset, For the old dataset, is a randomly generated interpolation factor, e ∈[0,1], is the minority class data point, is the majority class data point, 、 、 、 The unit is consistent with that of the corresponding reservoir environment data;

[0023] The particle swarm optimization PSO algorithm is Update particle velocity and position to optimize the parameters of the reservoir dynamic behavior model; is the particle number, =1,2,...,N; is the number of iterations; For the The particle in The speed of the iteration, in meters per second; For the The particle in The position of the iteration, in meters; For the The historical optimal position of each particle, in meters; is the global optimal position of the particle, in meters; is the inertia weight factor, dimensionless; and is the acceleration constant, = =2; and is a uniformly distributed random number, ∈[0,1], ∈[0,1];

[0024] The specific process of regularly using new data sets to verify and update the model is: using cross-validation and A / B testing methods to evaluate model performance, ensure the stability and accuracy of the model, and be able to adapt to changes in reservoir conditions.

[0025] Preferably, the digital intelligence application layer includes the following connected in sequence:

[0026] The data visualization module uses Tableau tools to build dynamic dashboards that display key reservoir performance indicators (KPIs) and forecast results in real time. The KPIs include at least the three-dimensional spatiotemporal distribution of wellhead pressure, liquid production, and water cut.

[0027] The multi-terminal interaction module is configured to: implement a responsive user interface through the React framework (i.e., reaction framework), adaptable to PC (i.e., personal computer) or mobile terminal; integrate the OAuth 2.0 (i.e., Open Authorization 2.0) protocol to implement multi-factor identity authentication; and implement data hierarchical access control based on the RBAC (i.e., role-based access control) model;

[0028] The feedback processing module includes: a data annotation subsystem that allows users to add anomalies to prediction results through touch or voice input; a parameter adjustment subsystem that provides sliding controls for users to adjust model input parameter thresholds; and a feedback analysis subsystem that uses the TF-IDF algorithm (i.e., term frequency-inverse document frequency algorithm) to parse user-entered text feedback.

[0029] The decision support module matches the current operating conditions with the historical case library to output a risk assessment report that includes recommendations for production increase measures and equipment maintenance plans.

[0030] Preferably, the trusted computing layer includes:

[0031] The source data protection module, connected to the data acquisition system in the physical layer, digitally signs and encrypts the collected reservoir environmental data to ensure its authenticity and security. Data is transmitted using the secure channel TLS 1.3 (i.e., Transport Layer Security 1.3) to prevent leakage or tampering during transmission. The digital signature is based on public key infrastructure, and the encryption is performed using the AES-256 encryption algorithm (i.e., the Advanced Encryption Standard with a 256-bit key length).

[0032] The data processing monitoring module, connected to the decision tree module and the time series database InfluxDB in the data processing layer, implements role-based access control; performs data audits, tracks and records all data access and processing activities; and desensitizes sensitive data to protect the privacy of personal and business information.

[0033] The model security assurance module, connected to the intelligent twin layer, encrypts all data used in the training and application of reservoir dynamic behavior models and implements access control to ensure data security during model training. Regular model verification and update audits are conducted to ensure the reliability of model prediction results and the transparency of model changes.

[0034] The user data verification module connected to the multi-terminal interaction module of the digital intelligence application layer ensures that the data displayed in the digital intelligence application layer has been strictly verified to ensure the authenticity and integrity of the data; it realizes the identity authentication and authorization of user operations, and ensures the security of data operations through a multi-factor authentication mechanism.

[0035] Preferably, the specific processing process of the decision tree module is as follows:

[0036] a) Receive reservoir environmental data of the physical entity layer, including at least temperature T∈[0,200]℃, pressure P∈[0,100]MPa, production Q∈[0,500]m 3 / d, chemical composition concentration vector C=[ ,…, ],in, to Represents the concentration of hydrocarbon components, non-hydrocarbon gases and aqueous phase ions in the reservoir fluid, in ppm or mol / L; preferably, the hydrocarbon components include methane to pentane, the non-hydrocarbon gases include CO2 and H2S, and the aqueous phase ions include Na + and Cl - ;

[0037] b)

[0038]

[0039] Where D is the dataset, A∈{T,P,Q,C} is the candidate feature, and V is the number of values ​​of feature A;

[0040] c) Classification rules: ;

[0041] d) Output the classification label set with timestamp τ { ,..., } to the data processing and monitoring module in the trusted computing layer for verification. The verified data is stored in InfluxDB, where τ is the time stamp in the ISO 8601 standard format. to Indicates different classification labels.

[0042] Preferably, the specific processing process of the Kalman filter module is as follows:

[0043] Receive data from the time series database InfluxDB;

[0044] Apply Kalman filter algorithm to optimize data fusion process; fusion process application ;in, is the posterior estimate, is a priori estimate, is the Kalman gain, dimensionless; is the observed value; is the observation model, dimensionless; 、 、 The unit is consistent with the unit of the corresponding reservoir environment data and is determined according to the specific application scenario;

[0045] Output data fusion results.

[0046] Preferably, the data compression module applies Brotli compression algorithm to compress the data fusion result output by the Kalman filter module. ;in, is the input reservoir monitoring data, i.e., the data fusion result output by the Kalman filter module; The unit of compressed output data is bytes or kilobytes.

[0047] A modeling method based on the oil reservoir digital twin modeling system includes the following steps:

[0048] Step 1: Construction of physical entity layer: responsible for collecting actual operation data of oil reservoir;

[0049] Step 2: Data processing layer construction: responsible for processing the data collected from the physical layer, including data classification, storage, fusion, compression and transmission;

[0050] Step 3: Intelligent twin layer construction: used to build a digital twin model of the reservoir and perform model prediction and optimization;

[0051] Step 4: Build the digital intelligence application layer: Display the results to end users;

[0052] Step 5: Build a trusted computing layer: Responsible for the data security and reliability of the entire reservoir digital twin modeling system based on trusted computing.

[0053] Preferably, the reservoir digital twin modeling method based on trusted computing includes the following steps:

[0054] Step 1: Physical layer construction:

[0055] The physical layer is responsible for collecting the actual operating data of the reservoir. According to the actual situation of the reservoir, multi-parameter sensors, acoustic sensors, and optical fiber sensors are deployed and connected to the data acquisition system;

[0056] Step 2: Data processing layer construction:

[0057] The data processing layer is responsible for processing the data collected from the physical entity layer;

[0058] First, a decision tree module is used to classify the data collected by the data acquisition system. The decision tree module decomposes the decision process layer by layer according to the attributes of the input data until the preset classification accuracy is achieved and verified by the trusted computing layer.

[0059] Then, the time series database InfluxDB is used to store and manage the classified data;

[0060] Then apply the Kalman filter module and use the Kalman filter algorithm to optimize the data fusion process;

[0061] Then use the data compression module to compress the data using the Brotli data compression algorithm;

[0062] Finally, the data transmission module uses the MQTT protocol to ensure efficient and reliable data transmission;

[0063] Step 3: Building the Intelligent Twin Layer

[0064] Use recurrent neural networks (RNNs) to build reservoir dynamic behavior models and learn and simulate complex reservoir characteristics in detail;

[0065] The synthetic minority oversampling technique (SMOTE) is used to solve the imbalance problem of training data in reservoir dynamic behavior models, thereby improving the prediction and generalization capabilities of reservoir dynamic behavior models.

[0066] Particle swarm optimization (PSO) algorithm is used to optimize the parameters of the reservoir dynamic behavior model to achieve adaptive learning of the reservoir dynamic behavior model.

[0067] Regularly verify and update the reservoir dynamic behavior model using new data sets to ensure its accuracy and stability;

[0068] Step 4: Build the digital intelligence application layer:

[0069] Use Tableau data visualization tools to visualize data and display key performance indicators (KPIs) and forecast results in real time. Implement a responsive user interface using the React framework and implement multi-level access control and identity authentication. Interact with displayed data. Use decision support tools to provide strategic recommendations to optimize oilfield output and efficiency. Enable the digital intelligence application layer to display results to end users.

[0070] Step 5: Building a trusted computing layer:

[0071] The trusted computing layer is responsible for the data security and reliability of the entire trusted computing-based reservoir digital twin modeling system. It includes the following four key interaction modules: a source data protection module connected to the data acquisition system in the physical entity layer; a data processing and monitoring module connected to the decision tree module and time series database InfluxDB in the data processing layer; a model security assurance module connected to the intelligent twin layer; and a user data verification module connected to the multi-terminal interaction module in the digital application layer.

[0072] The source data protection module performs digital signature and encryption processing on the collected reservoir environmental data. The digital signature is based on the public key infrastructure, and the encryption is performed using the AES-256 encryption algorithm to ensure its authenticity and security. Data transmission is carried out using the secure channel TLS 1.3 protocol to prevent interception or tampering during transmission.

[0073] A data processing monitoring module provides role-based access control. A data audit system is equipped to conduct data audits, track, and record all data access and processing activities to support future data traceability and compliance audits. Sensitive data is desensitized to protect the privacy of personal and business information.

[0074] The model security assurance module encrypts all data used in the training and application of reservoir dynamic behavior models, implements end-to-end encryption, and uses the AES-256 algorithm to ensure data security during storage and processing. It implements access control to ensure that only authorized users and systems can access model training data, ensuring data security during model training. Regular model verification and update audits are conducted to ensure the reliability of model prediction results and the transparency of model changes. Regular model performance evaluations are conducted to continuously monitor the model's prediction results and ensure the long-term stability and reliability of the model.

[0075] The user data verification module ensures that the data displayed in the digital application layer has been strictly verified to ensure the authenticity and integrity of the data; implements identity authentication and authorization of user operations, and ensures the security of data operations through a multi-factor authentication mechanism.

[0076] The technical effects of the present invention are:

[0077] This invention incorporates advanced end-to-end encryption and blockchain technologies to enhance data security and traceability, ensuring the safety and integrity of data during collection, transmission, and processing. Designed to process real-time data and dynamically adjust, it can adapt to rapidly changing reservoir environments, enabling more efficient real-time monitoring and decision support. By integrating recurrent neural networks, an advanced data processing technique, it not only enhances the depth and accuracy of data analysis but also enables real-time adaptation to environmental changes, optimizing model performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 This is an architectural diagram of the intelligent twin modeling system described in the present invention. DETAILED DESCRIPTION

[0079] Example 1

[0080] 1. A trusted computing-based digital twin modeling system for oil reservoirs includes a physical entity layer for sequentially transmitting data and collecting real-time field data from oil reservoirs; a data processing layer responsible for processing the collected data; an intelligent twin layer for constructing a digital twin model of the oil reservoir and performing model prediction and optimization; and a digital intelligence application layer for data visualization. The system also includes a trusted computing layer for ensuring full data security. The trusted computing layer is connected to the physical entity layer, data processing layer, intelligent twin layer, and digital intelligence application layer.

[0081] 2. The modeling method based on the above-mentioned reservoir digital twin modeling system includes the following steps:

[0082] Step 1: Construction of physical entity layer: responsible for collecting actual operation data of oil reservoir;

[0083] Step 2: Data processing layer construction: responsible for processing the data collected from the physical layer, including data classification, storage, fusion, compression and transmission;

[0084] Step 3: Intelligent twin layer construction: used to build a digital twin model of the reservoir and perform model prediction and optimization;

[0085] Step 4: Build the digital intelligence application layer: Display the results to end users;

[0086] Step 5: Build a trusted computing layer: Responsible for the data security and reliability of the entire reservoir digital twin modeling system based on trusted computing.

[0087] Example 2

[0088] Based on the above embodiment 1,

[0089] Preferably, the physical entity layer includes a multi-parameter sensor, an acoustic wave sensor, an optical fiber sensor and a data acquisition system for real-time acquisition of reservoir environmental data. The data acquisition system is connected to the multi-parameter sensor, the acoustic wave sensor and the optical fiber sensor to improve the monitoring accuracy of the temperature and pressure changes of the oil well and the surrounding formation; the multi-parameter sensor can at least measure the temperature, pressure, production and chemical composition concentration of the oil reservoir; the unit of temperature is ℃, the unit of pressure is MPa, and the unit of production is m 3 / d, the unit of chemical component concentration is ppm or mol / L; the data acquisition system is SCADA (i.e., data acquisition and supervisory control system).

[0090] The data processing layer includes:

[0091] A decision tree module for classifying the collected reservoir environmental data to realize classification of the collected reservoir environmental data;

[0092] InfluxDB, a time series database for storing reservoir environmental data classified by a decision tree module. This database supports real-time query, analysis, and visualization to enable effective monitoring and management of reservoirs.

[0093] a Kalman filter module for optimizing the data fusion process;

[0094] A data compression module that uses the Brotli data compression algorithm to reduce the bandwidth and space occupied by data during storage and transmission;

[0095] A data transmission module that uses the MQTT protocol to ensure efficient and reliable data transmission;

[0096] The decision tree module is connected to the data acquisition system, and the data transmission module is connected to the intelligent twin layer.

[0097] The intelligent twin layer constructs a reservoir dynamic behavior model through a recurrent neural network (RNN), uses the synthetic minority oversampling technique (SMOTE) to address the imbalance of training data in the reservoir dynamic behavior model, and uses the particle swarm optimization (PSO) algorithm to optimize the parameters of the reservoir dynamic behavior model to achieve adaptive learning of the model. The model is regularly verified and updated with new data sets to ensure its accuracy and stability.

[0098] The recurrent neural network RNN ​​constructs a reservoir dynamic behavior model by simulating and predicting the fluid dynamics and multiphase flow characteristics in the reservoir and capturing the evolution of the reservoir state over time; the fluid dynamics and multiphase flow characteristics include the changes in the oil-water interface, pressure and production, and the units are m, MPa, m 3 / d;

[0099] The synthetic minority oversampling technique SMOTE is Generate new data sets to address the insufficiency or imbalance of the training data sets and improve the model's prediction and generalization capabilities; For a new dataset, For the old dataset, is a randomly generated interpolation factor, e ∈[0,1], is the minority class data point, is the majority class data point, 、 、 、 The unit is consistent with that of the corresponding reservoir environment data;

[0100] The particle swarm optimization PSO algorithm is Update particle velocity and position to optimize the parameters of the reservoir dynamic behavior model; is the particle number, =1,2,...,N; is the number of iterations; For the The particle in The speed of the iteration, in meters per second; For the The particle in The position of the iteration, in meters; For the The historical optimal position of each particle, in meters; is the global optimal position of the particle, in meters; is the inertia weight factor, dimensionless; and is the acceleration constant, = =2; and is a uniformly distributed random number, ∈[0,1], ∈[0,1];

[0101] The specific process of regularly using new data sets to verify and update the model is: using cross-validation and A / B testing methods to evaluate model performance, ensure the stability and accuracy of the model, and be able to adapt to changes in reservoir conditions.

[0102] The digital intelligence application layer is the user interface layer that displays the results to the end user, including:

[0103] The data visualization module uses Tableau tools to build dynamic dashboards that display key reservoir performance indicators (KPIs) and forecast results in real time. The KPIs include at least the three-dimensional spatiotemporal distribution of wellhead pressure, liquid production, and water cut.

[0104] The multi-terminal interaction module is configured to: implement a responsive user interface through the React framework, adaptable to PC or mobile terminals; integrate the OAuth 2.0 protocol to implement multi-factor identity authentication; and implement data hierarchical access control based on the RBAC model;

[0105] The feedback processing module includes: a data annotation subsystem that allows users to add anomalies to prediction results through touch or voice input; a parameter adjustment subsystem that provides sliding controls for users to adjust model input parameter thresholds; and a feedback analysis subsystem that uses the TF-IDF algorithm to parse user-entered text feedback.

[0106] The decision support module matches current operating conditions with a historical case library to output a risk assessment report containing recommendations for production increase measures and equipment maintenance plans, thereby optimizing oilfield output and efficiency.

[0107] The trusted computing layer includes:

[0108] The source data protection module, connected to the data acquisition system in the physical entity layer, digitally signs and encrypts the collected reservoir environmental data to ensure its authenticity and security. The module uses the secure channel TLS 1.3 protocol for data transmission to prevent leakage or tampering during transmission. The digital signature is based on the public key infrastructure, and the encryption is performed using the AES-256 encryption algorithm.

[0109] The data processing monitoring module, connected to both the decision tree module and the time series database InfluxDB in the data processing layer, implements role-based access control; conducts data audits, tracks and records all data access and processing activities, including data access, modification, and deletion; and desensitizes sensitive data to protect the privacy of personal and business information.

[0110] The model security assurance module, connected to the intelligent twin layer, encrypts all data used in the training and application of reservoir dynamic behavior models and implements access control to ensure data security during model training. Regular model verification and update audits are conducted to ensure the reliability of model prediction results and the transparency of model changes.

[0111] The user data verification module connected to the multi-terminal interaction module of the digital intelligence application layer ensures that the data displayed in the digital intelligence application layer has been strictly verified to ensure the authenticity and integrity of the data; it realizes the identity authentication and authorization of user operations, and ensures the security of data operations through a multi-factor authentication mechanism.

[0112] The specific processing process of the decision tree module is as follows:

[0113] a) Receive reservoir environmental data of the physical entity layer, including at least temperature T∈[0,200]℃, pressure P∈[0,100]MPa, production Q∈[0,500]m 3 / d, chemical composition concentration vector C=[ ,…, ],in, to Represents the concentration of hydrocarbon components, non-hydrocarbon gases and aqueous phase ions in the reservoir fluid, in ppm or mol / L; preferably, the hydrocarbon components include methane to pentane, the non-hydrocarbon gases include CO2 and H2S, and the aqueous phase ions include Na + and Cl - ;

[0114] b)

[0115]

[0116] Where D is the dataset, A∈{T,P,Q,C} is the candidate feature, and V is the number of values ​​of feature A;

[0117] c) Classification rules: ;

[0118] d) Output the classification label set with timestamp τ { ,..., } to the data processing and monitoring module in the trusted computing layer for verification. The verified data is stored in InfluxDB, where τ is the time stamp in the ISO 8601 standard format. to Indicates different classification labels.

[0119] The specific processing process of the Kalman filter module is as follows:

[0120] Receive data from the time series database InfluxDB;

[0121] Apply Kalman filter algorithm to optimize data fusion process; fusion process application ;in, is the posterior estimate, is a priori estimate, is the Kalman gain, dimensionless; is the observed value; is the observation model, dimensionless; 、 、 The unit is consistent with the unit of the corresponding reservoir environment data and is determined according to the specific application scenario;

[0122] The output data fusion results provide more comprehensive and accurate reservoir status information, including fused reservoir pressure and temperature distribution maps, for more accurate reservoir behavior prediction and management.

[0123] The data compression module uses the Brotli compression algorithm to compress the data fusion result output by the Kalman filter module. ;in, is the input reservoir monitoring data, i.e., the data fusion result output by the Kalman filter module; To compress output data, the unit is bytes or kilobytes. This data is used to reduce the bandwidth and space occupied by data transmission in remote oil field environments with limited bandwidth, increase data transmission speed and reduce transmission costs.

[0124] Example 3

[0125] The modeling method performed by the intelligent twin modeling system according to the second embodiment includes the following steps:

[0126] Step 1: Physical layer construction:

[0127] The physical entity layer is responsible for collecting the actual operating data of the oil reservoir. Sensors are deployed according to the actual situation of the oil reservoir. Multi-parameter sensors, acoustic wave sensors, and fiber optic sensors are all connected to the data acquisition system. The multi-parameter sensors can at least measure the temperature, pressure, production and chemical composition concentration of the oil reservoir. The acoustic wave sensors are used to monitor the acoustic characteristics of the formation and help analyze the formation structure and fluid dynamics. The fiber optic sensors are used to monitor the temperature and pressure changes of the oil reservoir and surrounding formations in real time. The sensors have a high-precision measurement capability of ±0.1% and a wide temperature operating range of -40°C to +150°C. They are extremely sensitive to environmental changes and can adapt to extreme oilfield environments. The data acquisition system is SCADA. Preferably, based on the actual conditions of the reservoir, all sensors and data acquisition systems are equipped with protective measures, including anti-corrosion, dustproof, waterproof and anti-vibration designs, to ensure the long-term stable operation of the equipment in harsh environments. At the same time, the multi-parameter sensors, acoustic sensors, optical fiber sensors and data acquisition systems have self-diagnosis functions, which can detect their own working status in real time and automatically alarm when a fault occurs, while attempting preliminary fault recovery. Preferably, the data acquisition system has anti-interference characteristics to ensure the stability and accuracy of data in complex external environments. It is also equipped with wireless data transmission (LTE and 5G networks), satellite communication methods, and network time protocols to ensure that the data is accurately aligned in time and transmitted in real time and efficiently.

[0128] Step 2: Data processing layer construction:

[0129] The data processing layer is responsible for efficiently and accurately processing the data collected from the physical entity layer;

[0130] First, the decision tree module is used to classify the data collected by the data acquisition system. The decision tree module decomposes the decision process layer by layer according to the attributes of the input data until the preset classification accuracy is achieved, as follows:

[0131] a) Receive reservoir environmental data of the physical entity layer, including at least temperature T∈[0,200]℃, pressure P∈[0,100]MPa, production Q∈[0,500]m 3 / d, chemical composition concentration vector C=[ ,…, ],in, to Represents the concentration of hydrocarbon components, non-hydrocarbon gases and aqueous phase ions in the reservoir fluid, in ppm or mol / L; preferably, the hydrocarbon components include methane to pentane, the non-hydrocarbon gases include CO2 and H2S, and the aqueous phase ions include Na + and Cl - ;

[0132] b)

[0133]

[0134] Where D is the dataset, A∈{T,P,Q,C} is the candidate feature, and V is the number of values ​​of feature A;

[0135] c) Classification rules: ;

[0136] d) Output the classification label set with timestamp τ { ,..., } to the data processing and monitoring module in the trusted computing layer for verification. The verified data is stored in InfluxDB, where τ is the time stamp in the ISO 8601 standard format. to Indicates different classification labels;

[0137] In this embodiment, the decision tree module automatically classifies the data into normal operating data, abnormal data, and data requiring further analysis by analyzing the data such as temperature, pressure, output, and chemical component concentration;

[0138] Then, we use the time series database InfluxDB to store and manage the classified data, which optimizes data writing and query efficiency and is particularly suitable for processing frequently collected reservoir monitoring data.

[0139] Then apply the Kalman filter module and use the Kalman filter algorithm to optimize the data fusion process; the Kalman filter module combines information from multiple data sources through a prediction-correction cycle to optimize the estimation process, thereby improving the overall data quality. Preferably, the Kalman filter module receives data from the time series database InfluxDB, and the fusion process is applied ;in, is the posterior estimate, is a priori estimate, is the Kalman gain, dimensionless; is the observed value, is the observation model, dimensionless; 、 、 The units are consistent with those of the corresponding reservoir environmental data and are determined according to the specific application scenario. The data fusion results are then output, including pressure and production data for reservoir monitoring. This technology is particularly suitable for processing dynamic system data with noise, providing more comprehensive and accurate reservoir status information for more accurate reservoir behavior prediction and management.

[0140] Then, the data compression module is used to compress the data using the Brotli data compression algorithm, which reduces the bandwidth and space occupied by the data during storage and transmission. These data are used to reduce the bandwidth and space occupied by data transmission in remote oil field environments with limited bandwidth, thereby increasing data transmission speed and reducing transmission costs. Compression processing is based on ;in, is the input reservoir monitoring data, i.e., the data fusion result output by the Kalman filter module; For compressed output data, the unit is bytes or kilobytes;

[0141] Finally, the data transmission module uses the MQTT (Message Queuing Telemetry Transport) protocol to ensure efficient and reliable data transmission. MQTT is a lightweight publish / subscribe messaging protocol suitable for high-latency or unreliable networks with low bandwidth.

[0142] Step 3: Building the Intelligent Twin Layer

[0143] The intelligent twin layer is the core layer used for advanced modeling and analysis in this invention. It is responsible for building and maintaining the digital twin model of the reservoir and performing model-based prediction and optimization.

[0144] A recurrent neural network (RNN) is used to construct a reservoir dynamic behavior model. The RNN simulates and predicts the fluid dynamics and multiphase flow characteristics in the reservoir, capturing the evolution of the reservoir state over time to build the reservoir dynamic behavior model. The reservoir dynamic behavior model can learn and simulate the complex characteristics of the reservoir in detail, providing accurate predictions of the reservoir's production behavior, including changes in the oil-water interface, pressure, and production. Fluid dynamics and multiphase flow characteristics include changes in the oil-water interface, pressure, and production.

[0145] The synthetic minority oversampling technology SMOTE is used to solve the problem of imbalanced training data in the reservoir dynamic behavior model, improve the prediction and generalization capabilities of the reservoir dynamic behavior model, enhance the training effect of the model, and improve the model's prediction ability for rare events; the synthetic minority oversampling technology SMOTE is used to solve the problem of imbalanced training data in the reservoir dynamic behavior model, improve the prediction and generalization capabilities of the reservoir dynamic behavior model, enhance the training effect of the model, and improve the model's prediction ability for rare events; Generate new data sets to address the insufficiency or imbalance of the training data sets and improve the model's prediction and generalization capabilities; For a new dataset, For the old dataset, is a randomly generated interpolation factor, is the minority class data point, is the majority class data point, 、 、 、 The unit is consistent with that of the corresponding reservoir environment data;

[0146] The particle swarm optimization PSO algorithm is used to optimize the parameters of the reservoir dynamic behavior model to achieve adaptive learning of the reservoir dynamic behavior model; the ... Update particle velocity and position to optimize the parameters of the reservoir dynamic behavior model; is the particle number, =1,2,...,N; is the number of iterations; For the The particle in The speed of the iteration, in meters per second; For the The particle in The position of the iteration, in meters; For the The historical optimal position of each particle, in meters; is the global optimal position of the particle, in meters; is the inertia weight factor, dimensionless; and is the acceleration constant, = =2; and is a uniformly distributed random number, ∈[0,1], ∈[0,1];

[0147] New data sets are regularly used to validate and update the reservoir dynamic behavior model, and cross-validation and A / B testing methods are used to evaluate model performance to ensure the accuracy and stability of the reservoir dynamic behavior model and its ability to adapt to changes in reservoir conditions.

[0148] Step 4: Build the digital intelligence application layer:

[0149] The digital intelligence application layer is the user interface layer, responsible for presenting complex data analysis results to end users in an intuitive and easy-to-understand manner, thereby affecting the user's operating experience and decision-making efficiency;

[0150] The digital intelligence application layer includes the following connected in sequence:

[0151] The data visualization module uses Tableau tools to build dynamic dashboards that display key reservoir performance indicators (KPIs) and forecast results in real time. These KPIs include at least the three-dimensional spatiotemporal distribution of wellhead pressure, liquid production, and water cut, enabling technicians and managers to quickly understand the current status of the oilfield.

[0152] The multi-terminal interaction module is configured to: implement a responsive user interface through the React framework, adaptable to PC or mobile terminals; integrate the OAuth 2.0 protocol to implement multi-factor identity authentication; and implement data hierarchical access control based on the RBAC model;

[0153] The feedback processing module interacts with the displayed data and includes: a data annotation subsystem that allows users to add anomaly marks to prediction results through touch or voice input; a parameter adjustment subsystem that provides sliding controls for users to adjust model input parameter thresholds; and a feedback analysis subsystem that uses the TF-IDF algorithm to parse user-entered text feedback. This not only enhances the system's interactivity and adaptability, but also continuously optimizes data analysis and model adjustments through user feedback.

[0154] The decision support module matches current operating conditions with a historical case library to output a risk assessment report containing recommendations for production increase measures and equipment maintenance plans, thereby optimizing oilfield output and efficiency.

[0155] These comprehensive measures significantly improve user decision-making efficiency and operational experience, while ensuring a high level of system security and reliability. Preferably, the interface design incorporates virtual reality and augmented reality technologies to provide an immersive user experience, utilizing Unity Shader programming, an advanced graphics rendering technology, to optimize 3D visual effects and enhance the quality of user interaction and data presentation.

[0156] Step 5: Building a trusted computing layer:

[0157] The trusted computing layer is responsible for the data security and reliability of the entire trusted computing-based reservoir digital twin modeling system. It includes the following four key interaction modules: a source data protection module connected to the data acquisition system in the physical entity layer, a data processing and monitoring module connected to the decision tree module and time series database InfluxDB in the data processing layer, a model security assurance module connected to the intelligent twin layer, and a user data verification module connected to the multi-terminal interaction module in the digital application layer.

[0158] The source data protection module performs digital signature and encryption processing on the collected reservoir environmental data. The digital signature is based on the public key infrastructure, and each piece of data is signed to ensure data integrity and non-repudiation. The encryption process is performed using the AES-256 encryption algorithm to ensure its authenticity and security. Data transmission is carried out using the secure channel TLS 1.3 protocol. TLS 1.3 reduces the number of round trips during the data handshake process, thereby reducing the delay in connection establishment, improving the efficiency of data transmission, and enhancing the ability to resist man-in-the-middle attacks to prevent leakage or tampering during transmission.

[0159] The data processing monitoring module implements role-based access control, assigning corresponding data access permissions based on user roles to control the scope and level of data access. For example, only authorized data analysts can access raw data analysis tools, while senior managers can access comprehensive reports from the decision support system. A data audit system is equipped to conduct data audits, track and record all data access and processing activities, including data access, modification, and deletion. Each operation will record the operation time, operator, operation type, and detailed content to support future data traceability and compliance audits. Sensitive data is desensitized to protect the privacy of personal and business information. For example, personal identity information and business secrets are anonymized or disguised to prevent sensitive information from being leaked during data analysis and sharing.

[0160] The model security assurance module encrypts all data used in the training and application of reservoir dynamic behavior models, implements end-to-end encryption, and uses the AES-256 algorithm to ensure data security during storage and processing. Access control is implemented to ensure that only authorized users and systems can access model training data, ensuring data security during the model training process. Regular model verification and update audits are conducted to ensure the reliability of model prediction results and the transparency of model changes. Regular model performance evaluation is conducted to continuously monitor the model's prediction results through indicators such as precision, recall rate, and F1 score to ensure the long-term stability and reliability of the model.

[0161] The user data verification module ensures that the data displayed in the digital intelligence application layer has been strictly verified to ensure data authenticity and integrity, and prevent erroneous or tampered data from affecting decision-making. It also implements user operation authentication and authorization, ensuring the security of data operations through multi-factor authentication mechanisms. For example, the user authentication process is strengthened through passwords, biometrics, and dynamic tokens.

[0162] The four modules in the trusted computing layer significantly improve the overall security protection level of the reservoir digital twin modeling system. It not only enhances the system's protection capabilities, but also increases users' trust in the system, thereby supporting the efficiency and accuracy of reservoir management.

Claims

1. A digital and intelligent reservoir twin modeling system based on trusted computing, characterized by: It includes a physical entity layer for real-time data collection of reservoir field data, which transmits data in sequence; a data processing layer responsible for processing the collected data; an intelligent twin layer for building a digital twin model of the reservoir and performing model prediction and optimization; and a digital intelligence application layer for data visualization. It also includes a trusted computing layer for ensuring full data security. The trusted computing layer is connected to the physical entity layer, data processing layer, intelligent twin layer, and digital intelligence application layer. The physical entity layer includes a multi-parameter sensor, an acoustic wave sensor, an optical fiber sensor, and a data acquisition system for real-time collection of reservoir environmental data. The data acquisition system is connected to the multi-parameter sensor, the acoustic wave sensor, and the optical fiber sensor. The multi-parameter sensor can at least measure the temperature, pressure, production, and chemical composition concentration of the reservoir. The data processing layer includes: a decision tree module for classifying collected reservoir environmental data; InfluxDB, a time series database for storing reservoir environmental data classified by a decision tree module; a Kalman filter module for optimizing the data fusion process; A data compression module that uses the Brotli data compression algorithm to reduce the bandwidth and space occupied by data during storage and transmission; A data transmission module that uses the MQTT protocol to ensure efficient and reliable data transmission; The decision tree module is connected to the data acquisition system, and the data transmission module is connected to the intelligent twin layer; The digital intelligence application layer includes the following connected in sequence: The data visualization module uses Tableau tools to build dynamic dashboards that display key reservoir performance indicators (KPIs) and forecast results in real time. The KPIs include at least the three-dimensional spatiotemporal distribution of wellhead pressure, liquid production, and water cut. The multi-terminal interaction module is configured to: implement a responsive user interface through the React framework, adaptable to PC or mobile terminals; integrate the OAuth 2.0 protocol to implement multi-factor identity authentication; and implement data hierarchical access control based on the RBAC model; The feedback processing module includes: a data annotation subsystem that allows users to add anomalies to prediction results through touch or voice input; a parameter adjustment subsystem that provides sliding controls for users to adjust model input parameter thresholds; Feedback analysis subsystem, which uses the TF-IDF algorithm to parse the text feedback input by users; The decision support module matches the current operating conditions with the historical case library to output a risk assessment report containing recommendations for production increase measures and equipment maintenance plans; The trusted computing layer includes: The source data protection module, connected to the data acquisition system in the physical entity layer, digitally signs and encrypts the collected reservoir environmental data to ensure its authenticity and security. The module uses the secure channel TLS 1.3 protocol for data transmission to prevent leakage or tampering during transmission. The digital signature is based on the public key infrastructure, and the encryption is performed using the AES-256 encryption algorithm. The data processing monitoring module, connected to the decision tree module and the time series database InfluxDB in the data processing layer, implements role-based access control; performs data audits, tracks and records all data access and processing activities; and desensitizes sensitive data to protect the privacy of personal and business information. The model security assurance module, connected to the intelligent twin layer, encrypts all data used in the training and application of reservoir dynamic behavior models and implements access control to ensure data security during model training. Regular model verification and update audits are conducted to ensure the reliability of model prediction results and the transparency of model changes. The user data verification module connected to the multi-terminal interaction module of the digital intelligence application layer ensures that the data displayed in the digital intelligence application layer has been strictly verified to ensure the authenticity and integrity of the data; it realizes the identity authentication and authorization of user operations, and ensures the security of data operations through a multi-factor authentication mechanism.

2. The reservoir digital twin modeling system based on trusted computing according to claim 1 is characterized by: The intelligent twin layer constructs a reservoir dynamic behavior model through a recurrent neural network (RNN), uses the synthetic minority oversampling technique (SMOTE) to address the imbalance of training data in the reservoir dynamic behavior model, and uses the particle swarm optimization (PSO) algorithm to optimize the parameters of the reservoir dynamic behavior model to achieve adaptive learning of the model. The model is regularly verified and updated with new data sets to ensure its accuracy and stability. The recurrent neural network RNN ​​constructs a reservoir dynamic behavior model by simulating and predicting the fluid dynamics and multiphase flow characteristics in the reservoir and capturing the evolution of the reservoir state over time; the fluid dynamics and multiphase flow characteristics include the change of the oil-water interface, pressure and production, and the units are m, MPa, m 3 / d; The synthetic minority oversampling technique SMOTE is Generate new data sets to address the insufficiency or imbalance of the training data sets and improve the model's prediction and generalization capabilities; For a new dataset, For the old dataset, is a randomly generated interpolation factor, ε ∈[0,1], is the minority class data point, is the majority class data point, 、 、 、 The unit is consistent with that of the corresponding reservoir environment data; The particle swarm optimization PSO algorithm is Update particle velocity and position to optimize the parameters of the reservoir dynamic behavior model; is the particle number, =1,2,...,N; is the number of iterations; For the The particle in The speed of the iteration, in meters per second; For the The particle in The position of the iteration, in meters; For the The historical optimal position of each particle, in meters; is the global optimal position of the particle, in meters; is the inertia weight factor, dimensionless; and is the acceleration constant, = =2; and is a uniformly distributed random number, ∈[0,1], ∈[0,1] ; The specific process of regularly using new data sets to verify and update the model is: using cross-validation and A / B testing methods to evaluate model performance, ensure the stability and accuracy of the model, and be able to adapt to changes in reservoir conditions.

3. The reservoir digital twin modeling system based on trusted computing according to claim 2 is characterized by: The specific processing process of the decision tree module is as follows: a) Receive reservoir environmental data of the physical entity layer, including at least temperature T∈[0,200]℃, pressure P∈[0,100]MPa, production Q∈[0,500]m 3 / d, chemical composition concentration vector C=[ ,…, ],in, to Represents the concentration of hydrocarbon components, non-hydrocarbon gases and aqueous phase ions in reservoir fluids, in ppm or mol / L; b） ， ； Where D is the dataset, A∈{T,P,Q,C} is the candidate feature, and V is the number of values ​​of feature A; c) Classification rules: ; d) Output the classification label set with timestamp τ { ,..., } to the data processing and monitoring module in the trusted computing layer for verification. The verified data is stored in InfluxDB, where τ is the time stamp in the ISO 8601 standard format. to Indicates different classification labels.

4. The reservoir digital twin modeling system based on trusted computing according to claim 3 is characterized by: The specific processing process of the Kalman filter module is as follows: Receive data from the time series database InfluxDB; Apply Kalman filter algorithm to optimize data fusion process; fusion process application ;in, is the posterior estimate, is a priori estimate, is the Kalman gain, dimensionless; is the observed value; is the observation model, dimensionless; 、 、 The units are consistent with those of the corresponding reservoir environment data; Output data fusion results.

5. The reservoir digital twin modeling system based on trusted computing according to claim 4 is characterized by: The data compression module uses the Brotli compression algorithm to compress the data fusion result output by the Kalman filter module. ;in, is the input reservoir monitoring data, i.e., the data fusion result output by the Kalman filter module; The unit of compressed output data is bytes or kilobytes.

6. A modeling method based on the reservoir digital twin modeling system according to claim 1, comprising the following steps: Step 1: Construction of physical entity layer: responsible for collecting actual operation data of oil reservoir; Step 2: Data processing layer construction: responsible for processing the data collected from the physical layer, including data classification, storage, fusion, compression and transmission; Step 3: Intelligent twin layer construction: used to build a digital twin model of the reservoir and perform model prediction and optimization; Step 4: Build the digital intelligence application layer: Display the results to end users; Step 5: Build a trusted computing layer: Responsible for the data security and reliability of the entire reservoir digital twin modeling system based on trusted computing.

Images