Ai big data model and optical fiber sensing-based smart oil field system and control method

By deploying a distributed multi-parameter fiber optic sensing system in the oilfield and combining it with an AI big data model, the problem of systematic application of big data analysis and fiber optic sensing in the oilfield has been solved, enabling efficient monitoring and optimized development of oil and gas resources, and improving oil and gas production efficiency and management intelligence.

WO2026232896A1PCT designated stage Publication Date: 2026-11-12OPTICAL SCI & TECH (CHENGDU) LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/CN2026/077424
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-09
Filing Date
2026-02-06
Publication Date
2026-11-12

AI Technical Summary

Technical Problem

Existing technologies are difficult to effectively combine big data analytics and fiber optic sensing in oilfields, resulting in a lack of systematic smart oilfield solutions and low efficiency in oil and gas resource development.

Method used

By combining a distributed multi-parameter fiber optic sensing system with an AI big data model, a smart oilfield system is established through the comprehensive and three-dimensional deployment of fiber optic sensing systems on the ground and downhole in the oilfield area. This system, combined with the artificial intelligence big data model for data processing and analysis, enables precise monitoring and optimized development of oil and gas resources.

Benefits of technology

It has enabled high-precision monitoring and optimized development of oil and gas resources, improved oil and gas production efficiency, and enhanced the recovery rate and intelligent level of production management in oil and gas fields.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2026077424_12112026_PF_FP_ABST
    Figure CN2026077424_12112026_PF_FP_ABST
Patent Text Reader

Abstract

An AI big data model and optical fiber sensing-based smart oil field system and a control method. The system comprises a multi-parameter optical fiber sensing system (1) three-dimensionally arranged on the ground and downhole, an optical fiber sensing data processing and interpretation system (2), and an artificial intelligence big data mode-based smart oil field system; the multi-parameter optical fiber sensing system (1) comprises an armored distributed multi-parameter sensing optical cable (11) internally comprising straight armored optical cables (12), an armored spiral optical cable (13), an armored distributed three-component DAS optical cable (14), and a distributed multi-parameter multi-component composite optical fiber sensing modulation and demodulation instrument (15). By means of using an artificial intelligence big data model, a high-precision oil reservoir fluid model for an underground oil and gas resource, and the distribution characteristics of unused or residual or remaining underground oil and gas resources, a development and exploitation solution to an underground oil and gas resource reservoir is optimized and designed, optimal adjustment wells or infill wells are deployed, and production systems are optimized, so as to develop the unused or residual or remaining underground oil and gas resources, to increase the oil and gas production and improving the recovery rate of the entire oil and gas field.
Need to check novelty before this filing date? Find Prior Art

Description

Intelligent oilfield system and control method based on AI big data model and fiber optic sensing Technical Field

[0001] This invention belongs to the field of artificial intelligence big data model and fiber optic sensing technology, specifically involving a smart oilfield system and control method based on AI big data model and fiber optic sensing. Background Technology

[0002] Artificial Intelligence (AI) is a crucial driving force behind the new round of technological revolution and industrial transformation. It is a new technological science that studies and develops theories, methods, technologies, and application systems to simulate, extend, and expand human intelligence. AI is an interdisciplinary and emerging field based on computer science, integrating computer science, psychology, philosophy, and other disciplines. It aims to understand the essence of intelligence and produce new intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI is an important component of the discipline of intelligence; it attempts to understand the essence of intelligence and produce new intelligent machines that can react in a way similar to human intelligence. AI is a very broad science, encompassing robotics, speech recognition, image recognition, natural language processing, expert systems, machine learning, and computer vision.

[0003] Large AI models typically refer to a class of AI models with a large number of parameters, constructed from artificial neural networks. These models are usually trained on massive amounts of data using self-supervised or semi-supervised learning. Initially, large models primarily referred to Large Language Models (LLMs). With technological advancements, the concept has expanded to include large visual models, multimodal models, and basic scientific models. Large models are a relatively new concept, and currently, there is no absolutely authoritative definition. Therefore, there is no strict standard for the minimum number of parameters required for a large model. Currently, large models generally refer to models with tens of billions, hundreds of billions, or even trillions of parameters. Furthermore, it is also common to refer to models with billions of parameters that have been pre-trained on large-scale data (significantly more training data than traditional pre-trained models) as large models. A significant characteristic of large AI models is their massive number of parameters. The number of parameters refers to the sum of all trainable parameters in the model, which typically determines the model's capacity and learning ability. As the number of parameters increases, large models can capture more features and more complex patterns, thus exhibiting greater performance in handling complex data and learning high-dimensional relationships.

[0004] Large models refer to machine learning models with a massive number of parameters and complex computational structures. These models are typically built from deep neural networks and have billions or even hundreds of billions of parameters. The purpose of large models is to improve their expressive power and predictive performance, enabling them to handle more complex tasks and data. Large models have wide applications in various fields, including natural language processing, computer vision, speech recognition, and recommender systems. By training on massive amounts of data to learn complex patterns and features, large models possess stronger generalization capabilities and can make accurate predictions on unseen data.

[0005] Big data analytics refers to the process of processing, analyzing, and mining large volumes of high-speed, multi-source, and multi-type data to discover valuable information and knowledge. Core technologies of big data analytics include data storage, data processing, data mining, data analysis, and data visualization. With the development of technologies such as the Internet, artificial intelligence, and the Internet of Things, the scale and complexity of data are constantly increasing, making big data analytics an indispensable part of enterprises and organizations.

[0006] In the field of deep learning, model compression and deployment are important research topics, and model distillation is one effective method. Model distillation was first proposed by Hinton et al. in 2015. Its core idea is to transfer knowledge from a complex large model (teacher model) to a relatively simple small model (student model) through knowledge transfer. Simply put, it uses the prediction probability distribution of the teacher model as a soft label to train the student model, thereby greatly reducing the complexity and computational resource requirements of the model while maintaining high prediction performance, achieving lightweight and efficient modeling.

[0007] Large model distillation, simply put, is the process of transferring knowledge from a complex, large model (the teacher model) to a smaller model (the student model). It's like a teacher imparting their extensive knowledge to students, enabling them to demonstrate abilities similar to the teacher's, even with limited resources.

[0008] Large-scale model distillation draws on the concept of "knowledge transfer" in education, using soft labeling to transfer knowledge from the teacher's model to the student's model. The specific process is as follows:

[0009] Teacher model training: First, train a large teacher model to achieve a high performance level.

[0010] Knowledge transfer: The output of the teacher model (such as probability distribution, intermediate layer features, etc.) is used as soft labels to guide the learning of the student model. The soft labels not only contain information about the correct category, but also carry information about the relationships between categories, enabling the student model to learn knowledge more effectively.

[0011] Student model optimization: Through these soft labels, the student model can learn the decision logic and feature representation of the teacher model, thereby improving performance.

[0012] (1) Large-scale distillation method

[0013] Response-based distillation: The student model learns to mimic the output of the teacher model (predicting results only), while the teacher model is not trained during the distillation process.

[0014] Distillation based on intermediate layer features: This method not only mimics the output of the teacher model, but also the feature representation of its intermediate layers, which can better convey the internal knowledge of the teacher model.

[0015] Distillation based on loss function: By adjusting the loss function, the student model can better learn the knowledge of the teacher model.

[0016] (2) Advantages of large-scale model distillation

[0017] Reduced computational costs: The computational complexity of small models is much lower than that of large models. During the inference phase, small models can quickly provide results, which is especially important for applications with high real-time requirements, such as intelligent customer service and voice assistants.

[0018] Reduced storage requirements: Smaller models have fewer parameters and occupy less storage space, making them easier to deploy in resource-constrained environments such as mobile devices and embedded devices.

[0019] Improved training efficiency: Training small models requires less time and resources, enabling the training process to be completed faster and meeting the needs of rapid iteration.

[0020] (3) Application scenarios of large-scale model distillation

[0021] AI applications on mobile devices, such as smart photography and voice recognition on smartphones, can reduce device energy consumption and computational burden without affecting user experience through large-scale model distillation.

[0022] Edge computing: In IoT devices, due to the limited computing power of the devices, distilled small models can be used to process data directly on the edge devices, reducing data transmission latency and costs.

[0023] Online services: For online services that require processing a large number of user requests, such as search engines and recommendation systems, using small models can improve service response speed and increase user satisfaction.

[0024] (4) Challenges of large-scale model distillation

[0025] Knowledge loss: During the knowledge transfer process, the smaller model may not be able to fully learn all the knowledge of the teacher model, resulting in a decrease in performance.

[0026] Optimization of distillation algorithms: How to design more efficient distillation algorithms so that small models can better learn the knowledge of teacher models remains a research hotspot.

[0027] Model adaptation: Different tasks and datasets require different distillation strategies. How to choose the appropriate strategy so that the distilled model performs best on a specific task is also a problem that needs to be solved.

[0028] (5) Recent Developments in Large Model Distillation

[0029] DDK Framework: In 2024, researchers proposed the DDK (Distilling Domain Knowledge for Efficient Large Language Models) framework. This framework dynamically adjusts the composition of the distillation dataset based on the performance differences between teacher and student models in different domains, making the distillation process more stable and efficient.

[0030] Distilling Step-by-Step: In 2023, researchers proposed a new distillation mechanism called "Distilling Step-by-Step", which allows smaller task-specific models to be trained with less training data and outperforms large language models with few-sample cues.

[0031] Large-model distillation provides an effective approach to solving the application of large models in resource-constrained scenarios. With continuous technological development and improvement, large-model distillation is expected to play a significant role in more fields, promoting the widespread application of artificial intelligence technology.

[0032] The concept of a smart oilfield is entirely new and still in its developmental stage, lacking authoritative and unified definitions and standards. It evolves from the digital oilfield, representing a qualitative leap from quantitative change. The core of a digital oilfield is digitization and intelligentization, emphasizing artificial intelligence; a smart oilfield, building upon the foundation of a digital oilfield, integrates human wisdom, emphasizing the combination of artificial intelligence and human intelligence. Because objects possess intelligence, while humans possess wisdom, unlike digital oilfields which focus on data collection, smart oilfields place greater emphasis on data processing and in-depth application, transforming "data" into "knowledge." Based on this knowledge, they assist and guide oilfield production decisions, thereby optimizing traditional processes, providing scientific management methods, and achieving a leap from static to dynamic, from intelligent to wise, from simple to in-depth, and from passive to proactive. The core idea of ​​a smart oilfield is to fully utilize information technology to thoroughly perceive, comprehensively interconnect, deeply integrate, and effectively consolidate key information from the core systems of oilfield operation, and to intelligently respond to the needs of oilfield production, management, and residents' lives at all levels. This will provide oilfield managers with scientific and efficient management tools and residents in the mining area with a better quality of life.

[0033] 1. A smart oilfield is an oilfield connected by the Internet of Things (IoT). A smart oilfield is necessarily an oilfield where everything is interconnected. The Internet of Things is hailed as the third information industry revolution after computers and the Internet. It refers to a network system that uses sensing devices such as RFID, sensors, QR codes, GPS, and cameras, according to agreed communication protocols, to connect specific objects with information networks and centralized control systems for information exchange and automatic control, thereby achieving intelligent identification, perception, positioning, tracking, monitoring, and management. The widespread application of the Internet of Things in oilfields allows people to have a comprehensive understanding of things and events in the oilfield, thus enabling them to "strategize and win battles from afar."

[0034] 2. A smart oilfield is an information-sharing oilfield. Comprehensive, systematic, high-quality, and shareable information is the foundation of a smart oilfield. Information can only realize its maximum value through sharing. The more individuals participate in the same information processing and application, the faster the social or economic value of the information grows, and the higher the degree of information sharing.

[0035] Based on the full interconnection of various application systems, by utilizing technologies such as information fusion, cloud computing, and fuzzy recognition, regional collaboration and data sharing can be achieved. Through the analysis and processing of information and data, objective, essential, and comprehensive cognition and judgment can be realized, thereby achieving visualized, measurable, and intelligent management and control of oil fields.

[0036] Data fusion and sharing aims to improve the efficiency of data and information utilization while breaking down the unfavorable situation of each unit and professional department establishing its own "information islands" by connecting the dots, and establishing a scientific and effective smart oilfield information sharing mechanism.

[0037] 3. A smart oilfield is an application- and service-oriented oilfield. The core of a smart oilfield is to establish a new oilfield ecosystem supported by new tools and technologies, encompassing oilfield production, management, and residents' lives. Through changes in management concepts and methods, it transforms the economic development model, achieving a leapfrog development from a traditional oilfield to an emerging one. Therefore, the ultimate goal of a smart oilfield is to provide a completely new management approach for oilfield exploration and development, oil and gas production, operation and management, and mining area services. By applying new methods, it enhances the ubiquity, visualization, and intelligence of all aspects, ultimately promoting the green, environmentally friendly, and sustainable development of the oilfield.

[0038] In summary, a smart oilfield consists of four parts: smart geology, smart engineering, smart management, and smart people's livelihood. Summary of the Invention

[0039] To address the aforementioned technical problems, this invention first provides a smart oilfield system based on AI big data models and fiber optic sensing, comprising:

[0040] A distributed or quasi-distributed multi-parameter fiber optic sensing system that is fully and three-dimensionally deployed on the surface and underground in the oilfield area;

[0041] The processing and interpretation system processes, analyzes, statistically analyzes, models, and labels distributed multi-parameter fiber optic sensing data collected in real time from the ground and well by the multi-parameter fiber optic sensing system.

[0042] The smart oilfield system based on the artificial intelligence big data model uses all geological structure information, geophysical data, logging data, well logging data, petroleum engineering data, production data of oil and gas production wells, and injection data of injection wells in the oilfield area to train the smart oilfield artificial intelligence big data model. Then, the AI ​​data processing model of distributed multi-parameter fiber optic sensing data on the surface and in the wells is extracted from the artificial intelligence big data model.

[0043] It also includes artificial seismic sources.

[0044] Furthermore, the multi-parameter fiber optic sensing system includes an armored distributed multi-parameter sensing optical cable, which contains a straight armored optical cable, an armored spiral optical cable, an armored distributed three-component DAS optical cable, and a distributed multi-parameter multi-component composite fiber optic sensing composite modulation and demodulation instrument.

[0045] The straight armored optical cable contains a first high reflectivity single-mode fiber, a first high reflectivity multimode fiber, a microstructure fiber, or a continuous fiber grating.

[0046] The armored spiral optical cable contains a cylindrical elastomer, around which a second high-reflectivity single-mode fiber and a second high-reflectivity multimode fiber are spirally wound. The angle between the spirally wound second high-reflectivity single-mode fiber and the second high-reflectivity multimode fiber and the extension direction of the cylindrical elastomer is within a certain degree.

[0047] The armored distributed three-component DAS optical cable contains an elastomer with a square or triangular cross-section. Each face of the elastomer is fixed with a third high-reflectivity single-mode optical fiber distributed in a sinusoidal or triangular pattern. A straight third high-reflectivity single-mode optical fiber is embedded in the center of the elastomer along the extension direction of the elastomer.

[0048] Furthermore, the armored distributed multi-parameter sensing optical cable is buried in shallow trenches on the ground in the oilfield area, and is also deployed outside or inside the casing or inside the tubing of all oil and gas production wells and injection wells in the oilfield.

[0049] Armored distributed multi-parameter sensing optical cables are used for real-time monitoring and measurement of surface deformation data, three-component fluid noise data, three-component strain data, temperature data, three-component SDAS microseismic data, two-dimensional or three-dimensional three-component SDAS seismic data, and three-component time-lapse SDAS seismic data.

[0050] The straight armored optical cable is used for real-time monitoring and measurement of fluid noise data, strain data, temperature data, seismic data in the strata surrounding the wellbore, downhole microseismic data, and time-shift vertical ground profile data outside the underground casing or tubing.

[0051] The armored spiral optical cable is used for real-time monitoring and measurement of fluid noise data, three-component strain data, temperature data, three-component seismic data in the strata surrounding the wellbore, downhole three-component microseismic data, and three-component time-lapse vertical ground profile data outside the underground casing or tubing.

[0052] The armored distributed three-component DAS optical cable is used for real-time monitoring and measurement of distributed three-component seismic data, distributed three-component microseismic data, and distributed three-component time-lapsed vertical ground profile data in the strata surrounding the ground or wellbore.

[0053] Furthermore, the artificial seismic source is one of the following: explosive seismic source, controllable seismic source, hammer seismic source, gas explosion seismic source, electrical energy seismic source, electric spark seismic source, air gun seismic source, or plasma seismic source.

[0054] The artificial seismic sources are deployed according to the location requirements of the construction design. They can generate zero-offset VSPs near the wellhead, or non-zero-offset VSPs far from the wellhead, or Walkaway VSPs or variable-offset VSPs by gradually moving from the wellhead to both ends, or Walkaround VSPs by moving around the wellhead at different radii, or 3D VSPs or time-shifted VSPs by uniformly deploying them in a grid pattern around the wellhead, or 2D or 3D ground SDAS earthquakes by uniformly deploying them in a 2D or 3D pattern within the oilfield area.

[0055] This invention also provides a control method for the aforementioned smart oilfield system based on AI big data models and fiber optic sensing, comprising the following steps:

[0056] (a) In accordance with the design requirements for the construction of smart oilfields, shallow trenches are dug on the ground to bury armored distributed multi-parameter sensing optical cables, and armored distributed multi-parameter sensing optical cables are laid outside or inside the casing or inside or outside the tubing of all oil and gas production wells and injection wells in the oilfield.

[0057] (b) Connect the first, second, or third high-reflectivity single-mode fiber, or high-reflectivity single-mode fiber, or high-reflectivity single-mode fiber in the armored distributed multi-parameter sensing optical cable buried in the shallow trenches and deployed in the wells within the oilfield area to the DAS signal input terminal of the composite modem, connect the first or second high-reflectivity multi-mode fiber to the DTS signal input terminal of the composite modem, connect the first or second high-reflectivity single-mode fiber to the DSS signal input terminal of the composite modem, and connect the microstructure fiber or continuous fiber grating to the DPS signal input terminal of the composite modem.

[0058] (c) Connect the armored distributed three-component DAS optical cable inside the armored distributed multi-parameter sensing optical cable to the three-component DAS signal input terminal of the composite modulation and demodulation instrument.

[0059] (d) Establish a smart oilfield data collection, storage, backup, processing, analysis, statistics, interpretation, real-time display and control center in the center of the smart oilfield. Upload the multi-parameter and multi-component sensing data of the ground and downhole fiber optic sensing networks output by all composite modulation and demodulation instruments to the smart oilfield data center in real time through ground transmission optical cables. In addition, upload the data collected by all ground and downhole electronic sensors and various valves in the smart oilfield to the smart oilfield data center through optical cables or cables.

[0060] (e) Using all the geological structure information, geophysical data and well logging data collected in the oilfield area, the processing and interpretation results of well logging data, petroleum engineering data, production data of oil and gas production wells, and injection data of injection wells, train the artificial intelligence big data model of smart oilfield, and then extract the AI ​​data processing model of distributed multi-parameter fiber optic sensing data of the surface and well from the artificial intelligence big data model.

[0061] (f) Using the artificial intelligence big data model trained in step (e), construct a precise geological structure model and oil and gas resource reservoir model for the smart oilfield;

[0062] (g) Using the accurate underground geological structure model and oil and gas reservoir model obtained in step (f), as well as the rock physical parameters and pore fluid parameters in the reservoir, a high-precision reservoir fluid model of the underground oil and gas resources in the smart oilfield is established.

[0063] (h) Integrate the three-dimensional coordinates of all oil and gas production wells and fluid injection wells in the smart oilfield, as well as the fluid outflow and injection data, and the production profile and water and gas intake profile data into the high-precision reservoir fluid model of the underground oil and gas resources obtained in step (g).

[0064] (i) Utilize the high-precision reservoir fluid model of underground oil and gas resources after the fusion of the artificial intelligence big data model of smart oilfield and the method of step (h), as well as the distribution characteristics of underground untapped, residual or remaining oil and gas resources, to optimize the development and exploitation scheme of underground oil and gas resource reservoirs, deploy the best adjustment wells or infill wells, optimize the production system, maximize the development of underground untapped, residual or remaining oil and gas resources, increase the single-well oil and gas production of existing oil and gas production wells and improve the recovery rate of the entire oil and gas field.

[0065] (j) Synchronously collect multi-parameter and multi-component sensor data deployed in the smart oilfield surface and downhole fiber optic sensor network in real time and long term, and use the AI ​​data processing model of the distributed multi-parameter fiber optic sensor data of the surface and downhole to process the multi-parameter and multi-component sensor data of the surface and downhole fiber optic sensor network in real time.

[0066] (k) In the VSP data acquisition work area of ​​the smart oilfield, start one or more seismic sources to excite at each pre-designed seismic source point according to the construction design. At the same time, collect various time-shifted ground seismic and time-shifted VSP data together through the time-shifted ground seismic and time-shifted DAS-VSP data acquisition and processing computer workstation system, composite modulation and demodulation instrument, and armored spiral optical cable and / or armored distributed three-component DAS optical cable.

[0067] (l) Amplitude-preserving processing is performed on the time-shifted ground earthquake and time-shifted DAS-VSP data acquired in step (k), and various ground earthquake P-wave and S-wave data attributes and various VSP P-wave and S-wave data attributes are extracted from the obtained amplitude-preserved high-resolution ground earthquake P-wave and S-wave imaging data volumes and amplitude-preserved high-resolution VSP P-wave and S-wave imaging data volumes.

[0068] (m) Using the various time-shifted ground seismic P-wave and S-wave data attributes extracted in step (l) and various time-shifted VSP P-wave and S-wave data attributes, based on the single attribute data or several combined attribute data that are most sensitive to the comprehensive interpretation target or reservoir physical parameters, a three-dimensional characterization and fine depiction of the geological structure or oil and gas reservoir spatial distribution within a certain range below the surface and around the VSP data acquisition well is carried out. The distribution range and changes of oil and gas resources in the oil and gas reservoir around the well are tracked, the oil / water boundary or gas / water boundary or oil / gas boundary is determined and changed, and the oil and gas saturation in the oil and gas reservoir is calculated and changed. Thus, a comprehensive evaluation of the oil and gas resource potential around the well and the distribution range of oil and gas reservoirs and the dynamic changes related to exploitation is achieved.

[0069] (n) Use the change data of oil / water boundary, gas / water boundary or oil / gas boundary in the underground reservoir obtained in step (m) to update and adjust the high-precision reservoir fluid model of underground oil and gas resources obtained in step (h) in real time.

[0070] (o) Repeat steps (j) to (n) of the process processing step (n) to update the high-precision reservoir fluid model of underground oil and gas resources, further optimize the design of the development and exploitation scheme of underground oil and gas resource reservoirs, deploy new adjustment wells or infill wells, further optimize the production system, maximize the development of underground untapped or residual or remaining oil and gas resources, further increase the single-well oil and gas production of existing oil and gas production wells and improve the recovery rate of the entire oil and gas field.

[0071] (p) Through the above steps and means, we will develop and improve the construction of smart oilfield infrastructure. Under the intelligent management of the smart oilfield data collection, storage, backup, processing, analysis, statistics, interpretation, real-time display and control center, we will gradually realize the AI ​​artificial intelligence management of the intelligent sensing and intelligent control system deployed on the ground and underground of the smart oilfield, and truly realize the comprehensive intelligent operation and intelligent management of the entire oil and gas field production process.

[0072] (q) Make full use of information technology to thoroughly perceive, fully interconnect, deeply and intelligently integrate key information of the core oilfield operation system, and make intelligent responses to needs at all levels. Attached Figure Description

[0073] Figure 1 is a schematic diagram of the structure and data processing method of the smart oilfield system based on AI big data model and fiber optic sensing of the present invention.

[0074] Figure 2 is a schematic diagram of the structure of the ground and underground distributed or quasi-distributed multi-parameter fiber optic sensing system of the present invention. Detailed Implementation

[0075] To facilitate understanding of the present invention, a more detailed description is provided below with reference to the accompanying drawings and specific embodiments. Preferred embodiments of the invention are shown in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described in this specification. Rather, these embodiments are provided to provide a more thorough and complete understanding of the disclosure of the present invention. They are not intended to limit the invention, but are merely illustrative, and the advantages of the invention will become clearer and easier to understand by illustrating them.

[0076] Figure 1 shows a schematic diagram of the structure and data processing method of the smart oilfield system based on AI big data model and fiber optic sensing according to the present invention.

[0077] The present invention relates to a smart oilfield system based on AI big data models and fiber optic sensing, comprising a distributed or quasi-distributed multi-parameter fiber optic sensing system 1, which is deployed in a three-dimensional manner on the surface and downhole of the oilfield area; a processing and interpretation system 2 based on three-dimensional fiber optic sensing data on the surface and downhole of the oilfield area; and a smart oilfield system 3 based on an artificial intelligence big data model.

[0078] Figure 2 shows a schematic diagram of the structure of the ground and underground distributed or quasi-distributed multi-parameter fiber optic sensing system of the present invention.

[0079] The multi-parameter fiber optic sensing system 1 includes an armored distributed multi-parameter sensing optical cable 11 deployed on the ground in the oilfield area, which contains a straight armored optical cable 12, an armored spiral optical cable 13, an armored distributed three-component DAS optical cable 14, and a composite modulation and demodulation instrument 15 for distributed multi-parameter multi-component composite fiber optic sensing.

[0080] The straight armored optical cable (12) contains a first high reflectivity single-mode fiber (121), a first high reflectivity multimode fiber (122), a microstructure fiber (123), or a continuous fiber grating (124).

[0081] The armored spiral optical cable 13 contains a cylindrical elastomer 131, and a high second reflectance single-mode optical fiber 132 and a high reflectance multimode optical fiber 133 are spirally wound around the cylindrical elastomer 131. The angle between the spirally wound second high reflectance single-mode optical fiber 132 and the second high reflectance multimode optical fiber 133 and the extension direction of the cylindrical elastomer 131 is about 60 degrees.

[0082] The armored distributed three-component DAS optical cable 14 contains an elastic body 141 with a square or triangular cross-section. Each surface of the elastic body 141 is fixed with a third high reflectivity single-mode optical fiber 142 distributed in a sinusoidal or triangular pattern. A straight third high reflectivity single-mode optical fiber 142 is embedded in the center of the elastic body 141 along the extension direction of the elastic body 141.

[0083] The processing and interpretation system 2 includes processing, analyzing, statistically analyzing, modeling, and labeling all real-time acquired surface and downhole distributed multi-parameter fiber optic sensing data; the smart oilfield system 3 based on the artificial intelligence big data model includes training a smart oilfield artificial intelligence big data model 31 using all geological structure information, geophysical (seismic and non-seismic) data, well logging data, well logging data, petroleum engineering data, production data of oil and gas production wells, and injection data of injection wells in the oilfield area, and then extracting an AI data processing model 32 of surface and downhole distributed multi-parameter fiber optic sensing data from the artificial intelligence big data model 31.

[0084] The armored distributed multi-parameter sensing optical cable 11 is buried in shallow trenches on the surface of the oilfield area, and simultaneously deployed outside or inside the casing or tubing of all oil and gas production wells and injection wells within the oilfield. The armored distributed multi-parameter sensing optical cable 11 buried in the shallow trenches on the surface of the oilfield area is used for real-time monitoring and measurement of surface deformation data, three-component fluid noise data, three-component strain data, temperature data, three-component SDAS microseismic data, two-dimensional or three-dimensional three-component SDAS seismic data, and three-component time-lapse SDAS seismic data.

[0085] The straight armored optical cable 12 is used for real-time monitoring and measurement of fluid noise data, strain data, temperature data, seismic data in the formation surrounding the wellbore, downhole microseismic data, and time-lapse vertical surface profile data outside the underground casing or tubing; the armored spiral optical cable 13 is used for real-time monitoring and measurement of fluid noise data, three-component strain data, temperature data, three-component seismic data in the formation surrounding the wellbore, downhole three-component microseismic data, and three-component time-lapse vertical surface profile data outside the underground casing or tubing; the armored distributed three-component DAS optical cable 14 is used for real-time monitoring and measurement of distributed three-component seismic data, distributed three-component microseismic data, and distributed three-component time-lapse vertical surface profile data in the formation surrounding the surface or wellbore.

[0086] The ground is also equipped with artificial seismic sources 4, which are one of the following: explosive seismic sources, controllable seismic sources, hammer seismic sources, gas explosion seismic sources, electrical seismic sources, electric spark seismic sources, air gun seismic sources, or plasma seismic sources.

[0087] The aforementioned intelligent oilfield system based on AI big data models and fiber optic sensing includes an artificial seismic source 4 deployed according to the construction design requirements. This source can generate zero-offset VSPs near the wellhead, or non-zero-offset VSPs away from the wellhead, or gradually move from the wellhead towards both ends to generate Walkaway VSPs or variable-offset VSPs, or move around the wellhead at different radii to generate Walkaround VSPs, or uniformly deploy in a three-dimensional grid around the wellhead to generate three-dimensional VSPs or time-shifted (four-dimensional) VSPs, or uniformly deploy in a two-dimensional or three-dimensional grid within the oilfield area to generate two-dimensional, three-dimensional, or time-shifted (four-dimensional) ground SDAS earthquakes.

[0088] The control method for a smart oilfield system based on AI big data models and fiber optic sensing includes the following steps:

[0089] (a) In accordance with the design requirements for the construction of a smart oilfield, a shallow trench is dug on the ground to bury the armored distributed multi-parameter sensing optical cable 11. The armored distributed multi-parameter sensing optical cable 11 is laid outside or inside the casing or outside or inside the tubing in all oil and gas production wells and injection wells in the oilfield.

[0090] (b) Connect the first high reflectivity single-mode fiber 121, the second high reflectivity single-mode fiber 132, or the third high reflectivity single-mode fiber 142 in the armored distributed multi-parameter sensing optical cable 11 buried in the shallow trenches and deployed in the wells within the oilfield area to the DAS signal input terminal of the composite modulation and demodulation instrument 15; connect the first high reflectivity multi-mode fiber 122 or the second high reflectivity multi-mode fiber 133 to the DTS signal input terminal of the composite modulation and demodulation instrument 15; connect the first high reflectivity single-mode fiber 121 or the second high reflectivity single-mode fiber 132 to the DSS signal input terminal of the composite modulation and demodulation instrument 15; and connect the microstructure fiber 123 or the continuous fiber grating 124 to the DPS signal input terminal of the composite modulation and demodulation instrument 15.

[0091] (c) Connect the armored distributed three-component DAS optical cable 14 inside the armored distributed multi-parameter sensing optical cable 11 buried in the shallow trenches and deployed in the wells within the oilfield area to the three-component DAS signal input terminal of the composite modulation and demodulation instrument 15.

[0092] (d) Establish a smart oilfield data collection, storage, backup, processing, analysis, statistics, interpretation, real-time display and control center in the center of the smart oilfield. Upload the multi-parameter and multi-component sensing data of the ground and downhole fiber optic sensing networks output by all composite modulation and demodulation instruments 15 to the smart oilfield data center in real time through ground transmission optical cables. In addition, upload the data collected by all ground and downhole electronic sensors and various valves in the smart oilfield to the smart oilfield data center through optical cables or cables.

[0093] (e) Using all the geological structure information, geophysical (seismic and non-seismic) data, well logging data and well logging data processing and interpretation results, petroleum engineering data, oil and gas production well production data, injection well injection data, etc. collected in the oilfield area, train the artificial intelligence big data model 31 of the smart oilfield, and then extract the AI ​​data processing model 32 of the distributed multi-parameter fiber optic sensing data of the surface and well from the artificial intelligence big data model 31.

[0094] (f) Using the artificial intelligence big data model of the smart oilfield trained in step (e), construct an accurate underground geological structure model and oil and gas resource reservoir model of the smart oilfield.

[0095] (g) Using the accurate underground geological structure model and oil and gas reservoir model obtained in step (f), as well as the rock physical parameters and pore fluid parameters in the reservoir, a high-precision reservoir fluid model of the underground oil and gas resources in the smart oilfield is established.

[0096] (h) Integrate the three-dimensional coordinates of all oil and gas production wells and fluid injection wells in the smart oilfield, as well as the fluid outflow and injection data, and the production profile and water and gas intake profile data into the high-precision reservoir fluid model of the underground oil and gas resources obtained in step (g).

[0097] (i) Utilize the high-precision reservoir fluid model of underground oil and gas resources after the fusion of the artificial intelligence big data model 31 of the smart oilfield and the method of step (h), as well as the distribution characteristics of underground untapped, residual or remaining oil and gas resources, to optimize the development and exploitation scheme of underground oil and gas resource reservoirs, deploy the best adjustment wells or infill wells, optimize the production system, maximize the development of underground untapped, residual or remaining oil and gas resources, increase the single-well oil and gas production of existing oil and gas production wells and improve the recovery rate of the entire oil and gas field.

[0098] (j) Synchronously collect multi-parameter and multi-component sensing data from the smart oilfield surface and downhole fiber optic sensing network in real time and long term, and use the AI ​​data processing model 32 to process the multi-parameter and multi-component sensing data from the surface and downhole distributed multi-parameter fiber optic sensing network in real time.

[0099] (k) In the VSP data acquisition work area of ​​the smart oilfield, one or more seismic sources are activated according to the construction design to excite at each pre-designed seismic source point. At the same time, various time-shifted ground seismic and time-shifted VSP data are acquired together through the time-shifted ground seismic and time-shifted DAS-VSP data acquisition and processing computer workstation system, composite modulation and demodulation instrument 15, armored spiral optical cable 13 and / or armored distributed three-component DAS optical cable 14.

[0100] (l) Amplitude-preserving processing is performed on the time-shifted ground earthquake and time-shifted DAS-VSP data acquired in step (k), and various ground earthquake P-wave and S-wave data attributes and various VSP P-wave and S-wave data attributes are extracted from the obtained amplitude-preserved high-resolution ground earthquake P-wave and S-wave imaging data volumes and amplitude-preserved high-resolution VSP P-wave and S-wave imaging data volumes.

[0101] (m) Using the various time-shifted ground seismic P-wave and S-wave data attributes extracted in step (l) and various time-shifted VSP P-wave and S-wave data attributes, based on the single attribute data or several combined attribute data that are most sensitive to the comprehensive interpretation target or reservoir physical parameters, a three-dimensional characterization and fine depiction of the geological structure or oil and gas reservoir spatial distribution within a certain range below the surface and around the VSP data acquisition well is carried out. The distribution range and changes of oil and gas resources in the oil and gas reservoir around the well are tracked, the oil / water boundary or gas / water boundary or oil / gas boundary is determined and changed, and the oil and gas saturation in the oil and gas reservoir is calculated and changed. Thus, a comprehensive evaluation of the oil and gas resource potential around the well and the distribution range of oil and gas reservoirs and the dynamic changes related to exploitation is achieved.

[0102] (n) Use the change data of oil / water boundary, gas / water boundary or oil / gas boundary in the underground reservoir obtained in step (m) to update and adjust the high-precision reservoir fluid model of underground oil and gas resources obtained in step (h) in real time.

[0103] (o) Repeat steps (j) to (n) of the process processing step (n) to update the high-precision reservoir fluid model of underground oil and gas resources, further optimize the design of the development and exploitation scheme of underground oil and gas resource reservoirs, deploy new adjustment wells or infill wells, further optimize the production system, maximize the development of underground untapped or residual or remaining oil and gas resources, further increase the single-well oil and gas production of existing oil and gas production wells and improve the recovery rate of the entire oil and gas field.

[0104] (p) Through the above steps and means, we will develop and improve the construction of smart oilfield infrastructure. Under the intelligent management of the smart oilfield data collection, storage, backup, processing, analysis, statistics, interpretation, real-time display and control center, we will gradually realize the AI ​​artificial intelligence management of the intelligent sensing and intelligent control system deployed on the ground and underground of the smart oilfield, and truly realize the comprehensive intelligent operation and intelligent management of the entire oil and gas field production process.

[0105] (q) Make full use of information technology to thoroughly perceive, fully interconnect, deeply and intelligently integrate key information of the core oilfield operation system, and make intelligent responses to the needs of oilfield production, management and residents' lives at all levels, so as to provide oilfield managers with scientific and efficient management tools, provide mining area residents with a better quality of life, and build a truly smart oilfield system.

Claims

1. A smart oilfield system based on AI big data models and fiber optic sensing, characterized in that, include: A distributed or quasi-distributed multi-parameter fiber optic sensing system that is fully deployed in the oilfield area on the ground and downhole (1); The processing and interpretation system (2) processes, analyzes, statistically analyzes, models, and establishes tags for the distributed multi-parameter fiber optic sensing data collected in real time by the multi-parameter fiber optic sensing system (1) on the ground and in wells. The intelligent oilfield system (3) based on the artificial intelligence big data model uses all geological structure information, geophysical data, logging data, well logging data, petroleum engineering data, production data of oil and gas production wells, and injection data of injection wells in the oilfield area to train the intelligent oilfield artificial intelligence big data model (31), and then extracts the AI ​​data processing model (32) of the distributed multi-parameter fiber optic sensing data of the surface and well from the artificial intelligence big data model (31); It also includes artificial seismic sources (4).

2. The intelligent oilfield system based on AI big data model and fiber optic sensing according to claim 1, characterized in that, The multi-parameter fiber optic sensing system (1) includes an armored distributed multi-parameter sensing optical cable (11), which contains a straight armored optical cable (12), an armored spiral optical cable (13), an armored distributed three-component DAS optical cable (14), and a composite modulation and demodulation instrument (15) for distributed multi-parameter multi-component composite fiber optic sensing. The straight armored optical cable (12) contains a first high reflectivity single-mode fiber (121), a first high reflectivity multimode fiber (122), a microstructure fiber (123), or a continuous fiber grating (124). The armored spiral optical cable (13) contains a cylindrical elastomer (131), and a second high reflectivity single-mode optical fiber (132) and a second high reflectivity multimode optical fiber (133) are spirally wound around the cylindrical elastomer (131). The angle between the spirally wound second high reflectivity single-mode optical fiber (132) and the second high reflectivity multimode optical fiber (133) and the extension direction of the cylindrical elastomer (131) is 60 degrees. The armored distributed three-component DAS optical cable (14) contains an elastomer (141) with a square or triangular cross-section. Each surface of the elastomer (141) is fixed with a third high reflectivity single-mode optical fiber (142) distributed in a sinusoidal or triangular pattern. A straight third high reflectivity single-mode optical fiber (142) along the extension direction of the elastomer (141) is embedded in the center of the elastomer (141).

3. The intelligent oilfield system based on AI big data model and fiber optic sensing according to claim 2, characterized in that, The armored distributed multi-parameter sensing optical cable (11) is buried in a shallow trench on the ground in the oilfield area, and is also laid outside or inside the casing or inside the tubing of all oil and gas production wells and injection wells in the oilfield. Armored distributed multi-parameter sensing optical cable (11) is used for real-time monitoring and measurement of surface deformation data, three-component fluid noise data, three-component strain data, temperature data, three-component SDAS microseismic data, two-dimensional or three-dimensional three-component SDAS seismic data and three-component time-shifted SDAS seismic data. The straight armored optical cable (12) is used for real-time monitoring and measurement of fluid noise data, strain data, temperature data, seismic data in the strata surrounding the wellbore, downhole microseismic data, and time-shift vertical ground profile data outside the underground casing or pipe string. The armored spiral optical cable (13) is used for real-time monitoring and measurement of fluid noise data, three-component strain data, temperature data, three-component seismic data in the strata surrounding the wellbore, downhole three-component microseismic data and three-component time-lapse vertical ground profile data outside the underground casing or pipe string. The armored distributed three-component DAS optical cable (14) is used for real-time monitoring and measurement of distributed three-component seismic data, distributed three-component microseismic data and distributed three-component time-lapse vertical ground profile data in the strata surrounding the ground or wellbore.

4. The intelligent oilfield system based on AI big data model and fiber optic sensing according to claim 1, characterized in that, The artificial seismic source (4) is one of the following: explosive seismic source, controllable seismic source, hammer seismic source, gas explosion seismic source, electric energy seismic source, electric spark seismic source, air gun seismic source, or plasma seismic source.

5. The intelligent oilfield system based on AI big data model and fiber optic sensing according to claim 1, characterized in that, The artificial seismic source (4) is deployed according to the location requirements of the construction design. It can generate zero offset VSP near the wellhead, or generate non-zero offset VSP far away from the wellhead, or generate Walkaway VSP or variable offset VSP by gradually moving from the wellhead to both ends, or generate Walkaround VSP by moving around the wellhead at different radius distances, or generate three-dimensional VSP or time-shifted VSP by uniformly deploying it in a grid pattern around the wellhead, or generate two-dimensional or three-dimensional or time-shifted ground SDAS earthquakes by uniformly deploying it in a two-dimensional or three-dimensional pattern within the oilfield area.

6. A control method for a smart oilfield system based on AI big data models and fiber optic sensing as described in any one of claims 1 to 5, characterized in that, Includes the following steps: (a) In accordance with the design requirements for the construction of smart oilfield, armored distributed multi-parameter sensing optical cables (11) are buried in shallow trenches on the ground and armored distributed multi-parameter sensing optical cables (11) are laid outside or inside the casing or inside the tubing in all oil and gas production wells and injection wells in the oilfield. (b) Connect the first high reflectivity single-mode fiber (121), the second high reflectivity single-mode fiber (132), or the third high reflectivity single-mode fiber (142) in the armored distributed multi-parameter sensing optical cable (11) buried in the shallow trenches and deployed in the wells within the oilfield area to the DAS signal input end of the composite modulation and demodulation instrument (15), connect the first high reflectivity multi-mode fiber (122) or the second high reflectivity multi-mode fiber (133) to the DTS signal input end of the composite modulation and demodulation instrument (15), connect the first high reflectivity single-mode fiber (121) or the second high reflectivity single-mode fiber (132) to the DSS signal input end of the composite modulation and demodulation instrument (15), and connect the microstructure fiber (123) or continuous fiber grating (124) to the DPS signal input end of the composite modulation and demodulation instrument (15). (c) Connect the armored distributed three-component DAS optical cable (14) inside the armored distributed multi-parameter sensing optical cable (11) to the three-component DAS signal input terminal of the composite modulation and demodulation instrument (15). (d) Establish a smart oilfield data collection, storage, backup, processing, analysis, statistics, interpretation, real-time display and control center in the center of the smart oilfield. Upload the multi-parameter and multi-component sensing data of the ground and downhole fiber optic sensing networks output by all composite modulation and demodulation instruments (15) to the smart oilfield data center in real time through ground transmission optical cables. In addition, upload the data collected by all ground and downhole electronic sensors and various valves in the smart oilfield to the smart oilfield data center through optical cables or cables. (e) Using all the geological structure information, geophysical data and well logging data of the collected oilfield area, the processing and interpretation results of the well logging data, petroleum engineering data, production data of oil and gas production wells and injection data of injection wells, train the artificial intelligence big data model of smart oilfield (31), and then extract the AI ​​data processing model (32) of the distributed multi-parameter fiber optic sensing data of the surface and well from the artificial intelligence big data model (31). (f) Using the artificial intelligence big data model (31) trained in step (e), construct a precise geological structure model and oil and gas resource reservoir model for the smart oilfield; (g) Using the accurate underground geological structure model and oil and gas reservoir model obtained in step (f), as well as the rock physical parameters and pore fluid parameters in the reservoir, a high-precision reservoir fluid model of the underground oil and gas resources in the smart oilfield is established. (h) Integrate the three-dimensional coordinates of all oil and gas production wells and fluid injection wells in the smart oilfield, as well as the fluid outflow and injection data, and the production profile and water and gas intake profile data into the high-precision reservoir fluid model of the underground oil and gas resources obtained in step (g). (i) Using the high-precision reservoir fluid model of underground oil and gas resources after the fusion of the artificial intelligence big data model of smart oilfield (31) and the method of step (h), as well as the distribution characteristics of underground untapped or residual or remaining oil and gas resources, optimize the development and exploitation scheme of underground oil and gas resource reservoirs, deploy the best adjustment wells or infill wells, optimize the production system, maximize the development of underground untapped or residual or remaining oil and gas resources, increase the single-well oil and gas production of existing oil and gas production wells and improve the recovery rate of the entire oil and gas field; (j) Synchronously collect multi-parameter multi-component sensor data deployed in the smart oilfield surface and downhole fiber optic sensor network in real time and long-term, and use the AI ​​data processing model (32) of the distributed multi-parameter fiber optic sensor data of the surface and downhole to process the multi-parameter multi-component sensor data of the surface and downhole fiber optic sensor network in real time. (k) In the VSP data acquisition work area of ​​the smart oilfield, start one or more seismic sources to excite at each pre-designed seismic source point according to the construction design, and at the same time, collect various time-shifted ground seismic and time-shifted VSP data together through the time-shifted ground seismic and time-shifted DAS-VSP data acquisition and processing computer workstation system, composite modulation and demodulation instrument (15), armored spiral optical cable (13) and / or armored distributed three-component DAS optical cable (14); (l) Amplitude-preserving processing is performed on the time-shifted ground earthquake and time-shifted DAS-VSP data acquired in step (k), and various ground earthquake P-wave and S-wave data attributes and various VSP P-wave and S-wave data attributes are extracted from the obtained amplitude-preserved high-resolution ground earthquake P-wave and S-wave imaging data volumes and amplitude-preserved high-resolution VSP P-wave and S-wave imaging data volumes. (m) Using the various time-shifted ground seismic P-wave and S-wave data attributes extracted in step (l) and various time-shifted VSP P-wave and S-wave data attributes, based on the single attribute data or several combined attribute data that are most sensitive to the comprehensive interpretation target or reservoir physical parameters, a three-dimensional characterization and fine depiction of the geological structure or oil and gas reservoir spatial distribution within a certain range below the surface and around the VSP data acquisition well is carried out. The distribution range and changes of oil and gas resources in the oil and gas reservoir around the well are tracked, the oil / water boundary or gas / water boundary or oil / gas boundary is determined and changed, and the oil and gas saturation in the oil and gas reservoir is calculated and changed. Thus, a comprehensive evaluation of the oil and gas resource potential around the well and the distribution range of oil and gas reservoirs and the dynamic changes related to exploitation is achieved. (n) Use the change data of oil / water boundary, gas / water boundary or oil / gas boundary in the underground reservoir obtained in step (m) to update and adjust the high-precision reservoir fluid model of underground oil and gas resources obtained in step (h) in real time. (o) Repeat steps (j) to (n) of the process processing step (n) to update the high-precision reservoir fluid model of underground oil and gas resources, further optimize the design of the development and exploitation scheme of underground oil and gas resource reservoirs, deploy new adjustment wells or infill wells, further optimize the production system, maximize the development of underground unused or residual or remaining oil and gas resources, further increase the single-well oil and gas production of existing oil and gas production wells and improve the oil and gas recovery rate of the entire oil and gas field. (p) Through the above steps and means, we will develop and improve the construction of smart oilfield infrastructure. Under the intelligent management of the smart oilfield data collection, storage, backup, processing, analysis, statistics, interpretation, real-time display and control center, we will gradually realize the AI ​​artificial intelligence management of the intelligent sensing and intelligent control system deployed on the ground and underground of the smart oilfield, and truly realize the comprehensive intelligent operation and intelligent management of the entire oil and gas field production process. (q) Make full use of information technology to thoroughly perceive, fully interconnect, deeply and intelligently integrate key information of the core oilfield operation system, and make intelligent responses to needs at all levels.