A deep-sea high-temperature and high-pressure well drilling blowout risk evolution path identification method, system, processing device and storage medium

Abstract

This invention relates to a method, system, processing equipment, and storage medium for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling. The method includes: extracting key information from historical blowout accident reports, safety procedures, and expert experience documents of the deep-sea high-temperature and high-pressure drilling well under test; employing a multi-agent collaborative architecture to collect real-time operating data, equipment operation data, performance parameters, and geological environment data of the deep-sea high-temperature and high-pressure drilling well under test, and performing preprocessing to form a unified dataset; based on the extracted key information and the formed unified dataset, using a Transformer architecture finely tuned for the deep-sea drilling field, constructing a three-level blowout risk knowledge graph of the deep-sea high-temperature and high-pressure drilling well under test and dynamically updating it; and using a multi-agent collaborative architecture to traverse the three-level blowout risk knowledge graph, identifying the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling well under test. This invention can be widely applied in the field of deep-sea drilling safety monitoring technology.

Description

Technical Field

[0001] This invention relates to the field of deep-sea drilling safety monitoring technology, and in particular to a method, system, processing equipment, and storage medium for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling. Background Technology

[0002] Deep-sea high-temperature and high-pressure drilling operations face extremely complex geological and working conditions. Blowout risks are dynamically coupled by multiple risk factors such as drilling fluid performance degradation, wellbore pressure imbalance, equipment failure, and formation fluid intrusion. The evolution process is characterized by multiple paths, nonlinearity, and concealment.

[0003] Currently, relevant risk assessment methods mainly rely on ISM (Interpretive Structural Model) for basic information extraction and path sorting, but there are still the following technical defects: (1) Single data processing dimension: Existing ISM-based methods only focus on the structured extraction of historical accident data, and have insufficient ability to integrate multi-source heterogeneous data such as real-time operating data, equipment operation logs, and technical specification PDFs, making it difficult to uncover the implicit risk associations contained in unstructured texts; (2) Limitations in risk path identification: Relying solely on ISM analysis cannot systematically characterize the dynamic chain evolution relationship between risk elements, has weak ability to identify implicit transmission paths, and lacks quantitative assessment methods; (3) Insufficient model adaptability: Traditional ISM-derived models are mostly static structures, lack dynamic update mechanisms, and are difficult to adapt to the rapid changes in operating parameters during deep-sea high-temperature and high-pressure drilling, resulting in delayed risk warnings; (4) Lack of visualization and interactivity: Existing methods present risk paths in a single way, lacking multi-dimensional visualization design and interactive analysis functions, which is not conducive to on-site personnel quickly understanding the risk evolution law.

[0004] Although existing studies have attempted to optimize the identification of risk factors through ISM analysis, core technical challenges such as multi-source data fusion, dynamic path identification, quantitative assessment, and visualization have not yet been solved, which restricts the accuracy and timeliness of blowout risk early warning and fails to provide reliable safety guarantees for deep-sea high-temperature and high-pressure drilling operations. Summary of the Invention

[0005] To address the aforementioned problems, the purpose of this invention is to provide a method, system, processing equipment, and storage medium for identifying the evolution path of blowout risks in deep-sea high-temperature and high-pressure drilling, which can provide reliable safety assurance for deep-sea high-temperature and high-pressure drilling operations.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: Firstly, it provides a method for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling, comprising:

[0007] Key information was extracted from historical blowout accident reports, safety procedures, and expert experience documents of the deep-sea high-temperature and high-pressure drilling wells to be tested; A pre-built multi-agent collaborative architecture is used to collect real-time operating data, equipment operation data, performance parameters and geological environment data of the deep-sea high-temperature and high-pressure drilling well under test, and after preprocessing, a unified dataset is formed. Based on the extracted key information and the unified dataset, a three-level blowout risk knowledge graph of the deep-sea high-temperature and high-pressure drilling to be tested is constructed and dynamically updated using a Transformer architecture finely tuned for the deep-sea drilling field. Using a pre-built multi-agent collaborative architecture, the system traverses a three-level blowout risk knowledge graph to identify the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling to be tested, and then visualizes them.

[0008] Furthermore, the method also includes: Using a pre-built multi-agent collaborative architecture, a decision support report is generated based on the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling to be tested.

[0009] Furthermore, the multi-agent cooperative architecture includes: The data-aware intelligent agent is used to process real-time data and detect anomalies using a sliding window algorithm, with abnormal data triggering early warning signals. Semantic understanding agents are used to support natural language query interaction, deeply analyze unstructured text data, and extract key risk information based on fine-tuned large models. The path reasoning agent is used to traverse the three-level blowout risk knowledge graph using a hybrid traversal algorithm that combines depth-first search and breadth-first search, and to identify explicit and implicit risk paths. The quantitative analysis agent is used to determine the weights of risk factors using the entropy weight method and to calculate the posterior probability of the path using a Bayesian network, thereby quantifying the probability and uncertainty of risk occurrence.

[0010] Furthermore, based on the extracted key information and the resulting unified dataset, a three-level blowout risk knowledge graph for the deep-sea high-temperature and high-pressure well under test is constructed and dynamically updated using a Transformer architecture finely tuned for the deep-sea drilling field, including: Using a semantic understanding agent, based on the Transformer architecture fine-tuned in the field of deep-sea drilling and the extracted key information, a large model is pre-trained. Risk source entities, influencing factor entities, accident consequence entities, and the relationships between various entities are extracted from the unified dataset, and a fine-tuned large model is obtained. Using the ISM algorithm, based on the extracted key information and the extracted risk source entities, influencing factor entities, accident consequence entities, and the relationships between various entities, a three-level blowout risk knowledge graph is constructed, including a risk source layer, a transmission path layer, and an accident consequence layer. The three-level blowout risk knowledge graph is G(V, E, W), where V is the risk node, E is the edge, and W is the edge weight. By acquiring sensor and log data, and employing incremental learning algorithms to update the entity extraction and relationship recognition capabilities of the large model, incremental updates of the large model are achieved, thereby generating a dynamic knowledge graph that includes several types of risk nodes and related relationships.

[0011] Furthermore, the three-level blowout risk knowledge graph is a directed hierarchical network. The flow of risk is from bottom to top, that is, from the risk source layer to the transmission path layer, and finally converges to the accident consequence layer. The risk source layer includes the root causes of the risk, the transmission path layer includes the intermediate states and key events in the process of risk evolving from the source to the consequence, and the accident consequence layer includes the eventual undesirable events that may occur.

[0012] Furthermore, the method employs a pre-built multi-agent collaborative architecture to traverse a three-level blowout risk knowledge graph, identify the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling to be tested, and visualize them, including: A data-aware intelligent agent is used to continuously monitor real-time data streams. When an anomaly is detected, a path reasoning intelligent agent is used to traverse the three-level blowout risk knowledge graph and combine it with the attention mechanism of the large model to identify and filter all potential risk evolution paths, and select the core risk evolution paths that meet the requirements. A quantitative analysis agent was used to determine the weights of each risk factor in the three-level blowout risk knowledge graph and calculate the posterior probability of the core risk evolution path by combining the entropy weight method and Bayesian network, thus obtaining the quantitative analysis results. The ForceAtlas layout algorithm is used to construct a dual visualization mode of "knowledge graph + risk heat map" and the node styles are designed differently according to risk categories to visualize the evolution path of the selected core risks.

[0013] Furthermore, the aforementioned pre-built multi-agent collaborative architecture generates a decision support report based on the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling to be tested, including: Based on the heat value of risk nodes and the path correlation strength coefficient, set up multi-level early warning; Using a semantic understanding agent, a decision support report is generated based on visualization results, multi-level early warning settings, the core risk evolution path of the deep-sea high-temperature and high-pressure drilling to be tested, and quantitative analysis results.

[0014] Secondly, a system for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling is provided, including: The key information extraction module is used to extract key information from historical blowout accident reports, safety procedures, and expert experience documents of the deep-sea high-temperature and high-pressure drilling wells to be tested. The unified dataset construction module is used to collect real-time operating data, equipment operation data, performance parameters and geological environment data of the deep-sea high-temperature and high-pressure drilling to be tested, using a pre-built multi-agent collaborative architecture, and then preprocess the data to form a unified dataset. The knowledge graph construction module is used to construct and dynamically update a three-level blowout risk knowledge graph for the deep-sea high-temperature and high-pressure drilling to be tested, based on the extracted key information and the unified dataset formed, using a Transformer architecture finely tuned for the deep-sea drilling field. The core risk evolution path identification module is used to identify the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling to be tested by traversing the three-level blowout risk knowledge graph using a pre-built multi-agent collaborative architecture, and then visualize them.

[0015] Thirdly, a processing device is provided, including a computer program, wherein when the computer program is executed by the processing device, it is used to implement the steps corresponding to the above-mentioned deep-sea high-temperature and high-pressure drilling blowout risk evolution path identification method.

[0016] Fourthly, a computer-readable storage medium is provided, on which a computer program is stored, wherein the computer program, when executed by a processor, is used to implement the steps corresponding to the above-mentioned method for identifying the evolution path of deep-sea high-temperature and high-pressure drilling blowout risks.

[0017] The present invention has the following advantages due to the adoption of the above technical solutions: 1. This invention lays the foundation for a knowledge base through ISM analysis, and breaks through the limitations of traditional methods by integrating large model and knowledge graph technology. It achieves deep fusion of multi-source heterogeneous data, accurate identification of explicit and implicit risk paths, quantitative risk assessment and multi-dimensional visualization, which significantly improves the accuracy and operability of blowout risk early warning. It can be applied to the safety monitoring of deep-sea high-temperature and high-pressure drilling platforms, and provides key technical support for drilling operation safety.

[0018] 2. This invention uses a data-sensing agent and a semantic-understanding agent to collaborate, collecting real-time operating data at a frequency of 1-5Hz. At the same time, it parses unstructured texts such as historical accident reports and technical specification PDFs, extracts key information, and uniformly converts data in different formats such as CSV and Excel into JSON and establishes a standardized data dictionary, thereby achieving a unified representation of structured and unstructured data.

[0019] 3. This invention constructs a three-level blowout risk knowledge graph including a risk source layer, a transmission path layer, and an accident consequence layer. It uses a hybrid traversal algorithm combined with a large model attention mechanism to traverse the graph, which can accurately identify explicit and implicit risk paths.

[0020] 4. This invention uses an incremental learning algorithm to dynamically adjust the attributes and association strength of the graph entities every 1-10 minutes, which can achieve synchronous updates with drilling conditions. It uses the entropy weight method to determine the weight of risk factors and combines a Bayesian network to calculate the posterior probability of the path, thus possessing dynamic self-adaptation and real-time risk quantification capabilities.

[0021] 5. This invention uses the ForceAtlas layout algorithm to construct a dual visualization mode of "knowledge graph + risk heat map". Risk nodes are color-coded according to risk categories, which can realize multi-dimensional visualization and interactive analysis.

[0022] 6. This invention sets up a three-level early warning mechanism based on heat value and correlation strength, generates a decision support report, and forms intelligent decision support throughout the entire process.

[0023] In summary, this invention can be widely applied in the field of deep-sea drilling safety monitoring technology. Attached Figure Description

[0024] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. In the drawings: Figure 1 This is a schematic diagram of a method flow provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a knowledge graph of high-temperature and high-pressure drilling blowout risk provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the core evolution path of high-temperature and high-pressure drilling blowout risk provided by an embodiment of the present invention; Figure 4 This is a thermal schematic diagram of the risk of blowout in high-temperature and high-pressure drilling provided in an embodiment of the present invention. Detailed Implementation

[0025] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the invention and to fully convey the scope of the invention to those skilled in the art.

[0026] It should be understood that the terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “described” as used herein may also include the plural forms. The terms “comprising,” “including,” “containing,” and “having” are inclusive and therefore indicate the presence of the stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not construed as requiring them to be performed in a particular order described or illustrated unless the order of performance is explicitly indicated. It should also be understood that additional or alternative steps may be used.

[0027] Although terms such as first, second, third, etc., may be used in this document to describe multiple elements, components, regions, layers, and / or segments, these elements, components, regions, layers, and / or segments should not be limited by these terms. These terms may be used only to distinguish one element, component, region, layer, or segment from another. Unless the context clearly indicates otherwise, terms such as "first," "second," and other numerical terms used herein do not imply order or sequence. Therefore, the first element, component, region, layer, or segment discussed below may be referred to as the second element, component, region, layer, or segment without departing from the teachings of the exemplary embodiments.

[0028] Currently, although some studies have attempted to optimize the identification of risk factors through ISM analysis, core technical challenges such as multi-source data fusion, dynamic path identification, quantitative assessment, and visualization have not yet been solved. These challenges restrict the accuracy and timeliness of blowout risk warnings and fail to provide reliable safety guarantees for deep-sea high-temperature and high-pressure drilling operations. This invention provides a method for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling, comprising: extracting key information from historical blowout accident reports, safety procedures, and expert experience documents of the deep-sea high-temperature and high-pressure drilling well to be tested; using a pre-constructed multi-agent collaborative architecture, collecting real-time operating data, equipment operation data, performance parameters, and geological environment data of the deep-sea high-temperature and high-pressure drilling well to be tested, and performing preprocessing to form a unified dataset; based on the extracted key information and the formed unified dataset, using a Transformer architecture finely tuned for the deep-sea drilling field, constructing a three-level blowout risk knowledge graph of the deep-sea high-temperature and high-pressure drilling well to be tested and dynamically updating it; using the pre-constructed multi-agent collaborative architecture, traversing the three-level blowout risk knowledge graph, identifying the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling well to be tested, and visualizing them. This invention addresses the technical challenges of traditional risk assessment methods, such as weak data fusion capabilities, difficulty in identifying hidden paths, and low visualization levels, through five core modules: multi-source heterogeneous data acquisition and analysis, dynamic knowledge graph construction, intelligent reasoning of risk paths, multi-dimensional visualization presentation, and interactive analysis. It significantly improves the accuracy and operability of blowout risk early warning and provides key technical support for the safety of deep-sea high-temperature and high-pressure drilling operations.

[0029] Example 1 like Figure 1 As shown in the figure, this embodiment provides a method for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling, including the following steps: 1) Pre-construct a multi-agent collaborative architecture including a data-aware agent, a semantic-understanding agent, a path-reasoning agent, and a quantitative analysis agent, and realize data sharing and collaborative decision-making through agent communication protocols.

[0030] Specifically, the data-aware agent uses a sliding window algorithm to process real-time data and detect anomalies, triggering early warning signals based on abnormal data. The semantic understanding agent, based on a finely tuned large model, supports natural language query interaction, deeply analyzes unstructured text data, and extracts key risk information. The path reasoning agent uses a hybrid traversal algorithm combining depth-first search and breadth-first search to traverse the three-level blowout risk knowledge graph and leverages the large model's attention mechanism to focus on highly relevant nodes, accurately identifying explicit and implicit risk paths. The quantitative analysis agent uses entropy weighting to determine the weights of risk factors and combines this with a Bayesian network to calculate the posterior probability of the path, quantifying the likelihood and uncertainty of risk occurrence.

[0031] 2) Extract key information such as accident causes, failure modes, risk signs, and evolution stages from historical blowout accident reports, safety regulations, and expert experience documents of the deep-sea high-temperature and high-pressure drilling wells to be tested.

[0032] 3) Employing a data-sensing agent within a multi-agent collaborative architecture, real-time operating data, equipment operation data, performance parameters, and geological environment data of the deep-sea high-temperature and high-pressure drilling site are collected and preprocessed to form a unified dataset that integrates structured, semi-structured, and unstructured data. Specifically: 3.1) Real-time operating data such as drilling pressure, bottom hole temperature, drilling speed and displacement of the deep-sea high-temperature and high-pressure drilling to be tested are collected at a frequency of 1-5Hz through the drilling platform sensor network.

[0033] 3.2) Obtain equipment operation data of drilling equipment such as drill bits and blowout preventers, including sensor monitoring data and equipment fault records.

[0034] 3.3) Obtain performance parameters such as drilling fluid density, viscosity, and water loss rate.

[0035] 3.4) Obtain geological environment data such as formation pressure, porosity, and permeability.

[0036] 3.5) Obtain unstructured text data such as historical blowout accident reports, technical specification PDF documents, and expert experience and knowledge of the deep-sea high-temperature and high-pressure drilling well to be tested.

[0037] 3.6) Preprocess the collected and acquired data: 3.6.1) The box plot method and DBSCAN clustering algorithm are used to remove abnormal data in the real-time operating data, and missing values ​​are filled in by interpolation.

[0038] 3.6.2) Convert structured data in different formats such as CSV and Excel into a unified JSON format and establish a standardized data dictionary. The structured data includes equipment operation data, performance parameters and geological environment data.

[0039] 3.6.3) Perform word segmentation, part-of-speech tagging, and entity tagging on unstructured text data.

[0040] 3.6.4) Based on the preprocessed data, the annotation accuracy is optimized by combining the dictionary in the field of deep-sea drilling, forming a unified dataset that integrates structured, semi-structured and unstructured data.

[0041] 4) Based on the extracted key information and the resulting unified dataset, a three-level blowout risk knowledge graph for the deep-sea high-temperature and high-pressure drilling well under test is constructed and dynamically updated using a Transformer architecture finely tuned for the deep-sea drilling field. The three-level blowout risk knowledge graph includes a risk source layer, a transmission path layer, and an accident consequence layer, specifically: 4.1) A semantic understanding agent in a multi-agent collaborative architecture is adopted. Based on the Transformer architecture fine-tuned in the deep-sea drilling field and the extracted key information, a large model (such as BERT, GPT series) is pre-trained. Risk source entities (such as "insufficient drilling fluid density"), influencing factor entities (such as "temperature"), accident consequence entities (such as "blowout"), and the relationships between various entities are extracted from the unified dataset, and a fine-tuned large model is obtained.

[0042] Specifically, the Transformer architecture, fine-tuned for the deep-sea drilling field, utilizes labeled data derived from domain-specific texts such as historical blowout accident reports, safety regulations, and expert experience documents. This enables the Transformer architecture to acquire specialized capabilities for identifying entities and relationships related to deep-sea drilling blowout risks. Moreover, the fine-tuning goal of the Transformer architecture in this invention is not general language understanding, but rather the identification of specific risk elements, including accurately extracting risk entities such as "insufficient drilling fluid density," "gas intrusion," and "blowout preventer gate wear" from massive amounts of data, as well as risk transmission relationships such as "cause," "exacerbate," and "trigger."

[0043] Specifically, when using the Transformer architecture to extract entities and relationships, the extracted key information can be used as high-quality labeled data to fine-tune large models, enabling large models to more accurately identify specific entities (such as "insufficient drilling fluid density" and "gas intrusion") and their relationships (such as "cause" and "exacerbate") related to the risk of deep-sea drilling blowouts from massive amounts of data.

[0044] Specifically, the extracted risk source entities are used to construct the risk source layer of the three-level blowout risk knowledge graph, serving as the starting node for path identification and participating in weight calculation; the extracted influencing factor entities are used to construct the transmission path layer of the three-level blowout risk knowledge graph, serving as the intermediate state of risk transmission; the extracted accident consequence entities are used to construct the accident consequence layer of the three-level blowout risk knowledge graph, serving as the ending node for path identification and as the basis for the final risk level in the decision support report; and the extracted relationships between various entities are used to construct the edges of the three-level blowout risk knowledge graph, determining the direction of risk transmission and participating in the calculation of the association strength coefficient.

[0045] 4.2) Using the ISM algorithm, based on the extracted key information and the extracted risk source entities, influencing factor entities, accident consequence entities, and the relationships between various entities, a three-level blowout risk knowledge graph is constructed, including the risk source layer, transmission path layer, and accident consequence layer, to clearly present the hierarchical logic of risk evolution. The three-level blowout risk knowledge graph is G(V,E,W), where V is the risk node (the entity representation of risk factors or intermediate states), E is the edge (relationship), and W is the edge weight (relationship strength coefficient, initially determined by ISM and historical data).

[0046] Specifically, the three-level blowout risk knowledge graph is a directed hierarchical network. The flow of information (risk) is from bottom to top, that is, from the risk source layer to the transmission path layer, and finally converges to the accident consequence layer. The risk source layer (bottom layer) includes the root causes of the risk. For example: equipment factors: blowout preventer gate wear, sensor drift, mud pump failure; geological factors: abnormally high pressure formations, high permeability reservoirs, shallow gas; human or operational factors: drilling fluid density design too low, excessively fast tripping speed; environmental factors: high temperature at the bottom of the well, high pressure at the bottom of the well.

[0047] The transmission path layer (intermediate layer) includes the intermediate states and critical events in the process of risk evolving from its source to its consequences. For example: state changes: drilling fluid performance degradation, wellbore pressure fluctuations, formation fluid intrusion; critical events: well kick, overflow, circulation loss.

[0048] The accident consequences layer (top layer) includes the undesirable events that may ultimately occur. Examples include: blowout, uncontrolled blowout, platform fire, and explosion.

[0049] 4.3) Sensor and log data are acquired through a real-time data access interface. Incremental learning algorithms are used to update the entity extraction and relationship recognition capabilities of the large model. Entity attributes and association strength are dynamically adjusted every 1-10 minutes to achieve incremental updates of the large model, ensuring that the knowledge graph is synchronized with the actual drilling conditions. This generates a dynamic knowledge graph including several types of risk nodes and their relationships. 4.3.1) Data access: New sensor data (such as pressure and temperature) and log data (such as equipment fault alarms) are accessed in real time through message queues (such as Kafka).

[0050] 4.3.2) Feature extraction: The semantic understanding agent in the multi-agent collaborative architecture is used to perform feature engineering on new sensor data and log data, such as calculating the mean and variance of pressure over the past 10 minutes and extracting keywords from the logs.

[0051] 4.3.3) Incremental learning: ① Entity / Relationship Update: The features of new sensor data and log data are input into the fine-tuned large model. Without forgetting the original knowledge, the large model identifies new risk entities (such as "newly emerging vibration patterns") or new relationships (such as "vibration pattern A is strongly correlated with BHP fluctuations"). This can be achieved through online learning algorithms or incremental learning methods based on memory playback.

[0052] ② Attribute / Weight Update: The attributes (such as the real-time value of "drilling fluid density") and the correlation strength of edges (such as the correlation coefficient between "insufficient drilling fluid density" and "wellbore pressure drop") of existing nodes in the Level 3 blowout risk knowledge graph are recalculated and updated based on new sensor and log data, thereby generating a dynamic knowledge graph that includes several types of risk nodes and their relationships. For example, if the most recent 10 sets of data all show that both occur simultaneously, their correlation strength coefficient will increase.

[0053] Specifically, for example, a dynamic knowledge graph is generated that includes 12 types of risk nodes and 156 sets of relationships. The 12 types of risk nodes are: high temperature and high pressure mechanism, equipment failure or malfunction, abnormal drilling fluid performance, wellbore pressure imbalance, formation anomaly, operational error, monitoring and early warning failure, formation fluid intrusion, well kick or overflow, blowout preventer failure, blowout, and personnel response delay.

[0054] Specifically, the update cycle can be dynamically adjusted according to the drilling operation conditions.

[0055] 5) Employing a constructed multi-agent collaborative architecture, the system traverses a three-level blowout risk knowledge graph to identify the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling project under test, and visualizes them. Specifically: 5.1) The data-aware agent in the multi-agent collaborative architecture continuously monitors the real-time data stream. When an anomaly is detected (such as a sudden drop in bottom hole pressure), the path reasoning agent in the multi-agent collaborative architecture traverses the three-level blowout risk knowledge graph and, in conjunction with the attention mechanism of the large model, identifies and filters all potential risk evolution paths. The core risk evolution paths with a correlation strength coefficient ≥ 0.5 and a risk occurrence probability ≥ 30% are selected (more than 15 paths can be accurately identified). Among them, potential risk evolution paths include explicit risk paths and implicit correlation paths.

[0056] Specifically, the implementation process of the path reasoning agent includes: Input the three-level blowout risk knowledge graph G(V, E, W).

[0057] The hybrid traversal algorithm is divided into a breadth-first search (BFS) phase and a depth-first search (DFS) phase, wherein: ① Breadth-first search (BFS) phase: Starting from the active nodes marked by the data-aware agent (such as "bottom hole pressure anomaly"), a K-layer breadth-first search is performed. The purpose is to not miss any potentially affected neighboring nodes and short paths, and to ensure that all directly related risks are covered, i.e., to identify explicit paths.

[0058] ② Depth-First Search (DFS) Phase: Within the scope defined by BFS, a depth-first search is performed on the "high-risk" subgraph using the attention mechanism of a large model (such as BERT). The attention mechanism calculates the semantic similarity between the current node and key nodes in historical blowout accident cases, guiding the DFS algorithm to prioritize exploring paths that have historically led to serious consequences or are semantically highly matched with accident patterns, thereby discovering hidden paths.

[0059] Finally, path selection is performed: for all traversed paths, the risk score of each path is calculated by comprehensively considering the real-time popularity values ​​of all nodes on the path (provided by the quantitative analysis agent) and the correlation strength coefficient of all edges. The top N paths with the highest scores are then selected as the core risk evolution paths.

[0060] Specifically, the risk evolution path is the path formed by connecting the risk nodes in the three-level blowout risk knowledge graph through corresponding edges.

[0061] 5.2) Using a quantitative analysis agent in a multi-agent collaborative architecture, based on a combination of entropy weight method and Bayesian network, the weights of each risk factor in the three-level blowout risk knowledge graph are determined and the posterior probability of the core risk evolution path is calculated to obtain the quantitative analysis results. The occurrence probability of risk nodes in the three-level blowout risk knowledge graph is mapped to a heat value of 0-100, and the thickness of the edges represents the correlation strength coefficient (0-1 interval).

[0062] 5.3) The ForceAtlas layout algorithm is used to construct a dual visualization mode of "knowledge graph + risk heat map". The node styles are designed differently according to the risk categories (for example: risk nodes of high temperature and high pressure mechanism are in low saturation orange, risk nodes of equipment failure or malfunction are in gray, risk nodes of abnormal drilling fluid performance are in blue, risk nodes of wellbore pressure imbalance are in yellow, risk nodes of formation fluid invasion or well kick are in red, and risk nodes of operation or management error are in purple). The core risk evolution path is highlighted with a 3.5 pixel dark red line. The risk heat map uses color gradient to represent the risk level and clearly reflects the risk transmission characteristics.

[0063] Specifically, the process of constructing a risk heatmap includes: ① Based on historical and real-time data, calculate the probability of occurrence of each risk node and map the probability value to a heat value (0-100). The higher the heat value, the higher the risk level.

[0064] Specifically, historical data is used to construct the initial three-level blowout risk knowledge graph and train the large model. This includes key information extracted in step 2), as well as sensor data and equipment logs from historical drilling operations, covering various operating conditions and events. Historical data is used to calculate the prior probability of the risk and the baseline correlation strength.

[0065] Real-time data refers to the real-time operating condition data, equipment operation data, performance parameters, and geological environment data continuously collected in step 3) above during the current drilling operation to be tested. Real-time data is used to dynamically update the posterior probability of risk nodes and adjust the real-time correlation strength of edges, thereby realizing dynamic quantitative assessment of risk.

[0066] ② The ForceAtlas layout algorithm is used to visualize the layout of the knowledge graph. The width of the edge is positively correlated with the association strength coefficient (0-1). The association strength coefficient is determined by the correlation analysis between the historical accident transmission frequency and real-time data.

[0067] 6) Employing a semantic understanding agent within a multi-agent collaborative architecture, a decision support report is generated based on the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling operation under test. Specifically: 6.1) Based on the heat value of risk nodes and the path correlation strength coefficient, a three-level early warning system is set. Among them, the heat value of 30-50 and the correlation strength coefficient of 0.3-0.5 is a general early warning, which suggests paying attention to parameter changes; the heat value of 50-80 and the correlation strength coefficient of 0.5-0.8 is an important early warning, which suggests taking targeted prevention and control measures; the heat value of 80 or above and the correlation strength coefficient of 0.8 or above is an emergency early warning, which triggers the emergency response process.

[0068] 6.2) Employing a semantic understanding agent within a multi-agent collaborative architecture, this system generates a decision support report based on visualization results, a three-level early warning system, the core risk evolution path of the deep-sea high-temperature and high-pressure drilling test, and quantitative analysis results. The report includes risk level, key triggers, transmission paths, and emergency response recommendations. It supports exporting to multiple formats such as PDF and Word. Through an interactive interface, it supports node drag-and-drop adjustment (response time 0.5 seconds), risk path query, and other operations. It also supports seamless integration with the drilling control system.

[0069] Specifically, when a user enters a natural language question through an interactive interface, such as "What impact does high temperature have on drilling fluid?", the semantic understanding agent analyzes the question, transforms it into a graph query language, and retrieves relevant information from the knowledge graph.

[0070] The following detailed embodiments illustrate the method for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling according to the present invention: Deployment environment: A deep-sea drilling platform (drilling depth 5000m, bottom hole temperature 180℃, formation pressure coefficient 2.0), the system runs on an industrial control computer (CPU i7-12700K, memory 32GB, graphics card RTX 3060), operating system Windows 10 Professional.

[0071] Data Acquisition: Based on key information dimensions identified through ISM analysis, PDF documents such as the "Summary Report on the Gulf of Mexico Blowout Accident" and the "Technical Specifications for Deep-Sea High-Temperature and High-Pressure Drilling" were imported, and 42 relevant text entries were extracted. Sensor data for drilling pressure, bottom hole temperature, drilling speed, and drilling fluid density were also collected at a frequency of 2Hz.

[0072] Data preprocessing: Three abnormal data points were removed, missing data were filled in, and text annotation and format standardization were completed, with a text annotation accuracy rate of 92%.

[0073] Knowledge Graph Construction: Based on ISM analysis results and preprocessed data, a dynamic three-level blowout risk knowledge graph was generated, including 82 risk nodes and 156 edges. Figure 2 As shown, there are 8 nodes related to the high temperature and high pressure mechanism, 8 nodes related to the underbalanced condition, and 15 core path edges.

[0074] Path identification: Employing a constructed multi-agent collaborative architecture, the system traverses a three-tiered risk knowledge graph, identifying 48 potential risk evolution paths. After weighted filtering, the top 15 core risk evolution paths are determined, such as... Figure 3 As shown, the two core risk evolution paths with the highest weights are as follows: High pressure mechanism → increased pore pressure → formation pressure higher than wellbore pressure → underbalanced state → formation fluid invasion → well kick → blowout risk (path weight: 12.8).

[0075] High temperature action mechanism → thermal expansion effect → drilling fluid density change → insufficient drilling fluid density → drilling fluid failure → underbalanced state → formation fluid invasion → well kick → blowout risk (path weight: 11.5).

[0076] Knowledge graph visualization: Nodes related to the mechanism of high temperature and high pressure are displayed in low-saturation orange, and nodes related to detailed underequilibrium conditions are displayed in low-saturation pink; the core node for blowout risk is set to 1800 pixels in size, and the connecting line width of the core risk path is 3.5 pixels, marked in dark red.

[0077] Risk Heatmap: The risk propagation intensity between the high-pressure mechanism and the increase in pore pressure is 0.8, which is presented in dark red on the heatmap, as shown below. Figure 4 As shown.

[0078] Interactive operation: Supports custom adjustment of core node positions. Input the node identifier "underbalanced state" and the system will display its attribute information in real time (node ​​type is "risk condition", importance level 9; the predecessor associated node is "formation pressure is higher than wellbore pressure", and the successor associated node is "formation fluid intrusion", etc.).

[0079] Decision support report generation: Automatically outputs a decision support report in Word format, including key factors such as network density of 0.046, average risk path length of 6.2, and degree centrality of the core node "blowout risk" of 0.82. It also provides emergency response suggestions such as "strengthen real-time monitoring and dynamic adjustment of drilling fluid performance to maintain the stability of key parameters such as density and viscosity".

[0080] Example 2 This embodiment provides a system for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling, including: The key information extraction module is used to extract key information from historical blowout accident reports, safety procedures, and expert experience documents of the deep-sea high-temperature and high-pressure drilling wells to be tested. The unified dataset construction module is used to collect real-time operating data, equipment operation data, performance parameters and geological environment data of the deep-sea high-temperature and high-pressure drilling to be tested, using a pre-built multi-agent collaborative architecture, and then preprocess the data to form a unified dataset. The knowledge graph construction module is used to construct and dynamically update a three-level blowout risk knowledge graph for the deep-sea high-temperature and high-pressure drilling to be tested, based on the extracted key information and the unified dataset formed, using a Transformer architecture finely tuned for the deep-sea drilling field. The core risk evolution path identification module is used to identify the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling to be tested by traversing the three-level blowout risk knowledge graph using a pre-built multi-agent collaborative architecture, and then visualize them.

[0081] The system provided in this embodiment is used to execute the above-described method embodiments. For specific processes and details, please refer to the above embodiments, which will not be repeated here.

[0082] Example 3 This embodiment provides a processing device corresponding to the deep-sea high-temperature and high-pressure drilling blowout risk evolution path identification method provided in Embodiment 1. The processing device can be applied to client processing devices, such as mobile phones, laptops, tablets, desktop computers, etc., to execute the method of Embodiment 1.

[0083] The processing device includes a processor, a memory, a communication interface, and a bus. The processor, memory, and communication interface are connected via the bus to enable communication between them. The memory stores a computer program that can run on the processing device. When the processing device runs the computer program, it executes the deep-sea high-temperature and high-pressure drilling blowout risk evolution path identification method provided in Embodiment 1.

[0084] In some implementations, the memory may be high-speed random access memory (RAM), and may also include non-volatile memory, such as at least one disk storage device.

[0085] In other implementations, the processor can be any type of general-purpose processor, such as a central processing unit (CPU) or a digital signal processor (DSP), and there is no limitation here.

[0086] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0087] Those skilled in the art will understand that the structure of the above-described computing device is only a partial structure related to the present invention and does not constitute a limitation on the computing device to which the present invention is applied. A specific computing device may include more or fewer components, or combine certain components, or have different component arrangements.

[0088] Example 4 This embodiment provides a computer program product corresponding to the deep-sea high-temperature and high-pressure drilling blowout risk evolution path identification method provided in Embodiment 1. The computer program product may include a computer-readable storage medium on which computer-readable program instructions for executing the deep-sea high-temperature and high-pressure drilling blowout risk evolution path identification method described in Embodiment 1 are loaded.

[0089] A computer-readable storage medium can be a tangible device that holds and stores instructions for use by an instruction execution device. A computer-readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any combination thereof.

[0090] The computer-readable storage medium provided in the above embodiments has a similar implementation principle and technical effect to the above method embodiments, and will not be described again here.

[0091] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by a computer program. These computer programs can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that instructions executable by the processor of the computer or other programmable data processing apparatus generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0092] These computer programs may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0093] These computer programs may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0094] The above embodiments are only used to illustrate the present invention. The structure, connection method and manufacturing process of each component can be varied. All equivalent transformations and improvements made on the basis of the technical solution of the present invention should not be excluded from the protection scope of the present invention.

Claims

1. A method for identifying the evolution path of a deep-sea high-temperature high-pressure well drilling blowout risk, characterized in that, include: Key information was extracted from historical blowout accident reports, safety procedures, and expert experience documents of the deep-sea high-temperature and high-pressure drilling wells to be tested; A pre-built multi-agent collaborative architecture is used to collect real-time operating data, equipment operation data, performance parameters and geological environment data of the deep-sea high-temperature and high-pressure drilling well under test, and after preprocessing, a unified dataset is formed. Based on the extracted key information and the unified dataset, a three-level blowout risk knowledge graph of the deep-sea high-temperature and high-pressure drilling to be tested is constructed and dynamically updated using a Transformer architecture finely tuned for the deep-sea drilling field. Using a pre-built multi-agent collaborative architecture, the system traverses a three-level blowout risk knowledge graph to identify the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling to be tested, and then visualizes them.

2. The method for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling as described in claim 1, characterized in that, The method also includes: Using a pre-built multi-agent collaborative architecture, a decision support report is generated based on the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling to be tested.

3. The method for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling as described in claim 2, characterized in that, The multi-agent collaborative architecture includes: The data-aware intelligent agent is used to process real-time data and detect anomalies using a sliding window algorithm, with abnormal data triggering early warning signals. Semantic understanding agents are used to support natural language query interaction, deeply analyze unstructured text data, and extract key risk information based on fine-tuned large models. The path reasoning agent is used to traverse the three-level blowout risk knowledge graph using a hybrid traversal algorithm that combines depth-first search and breadth-first search, and to identify explicit and implicit risk paths. The quantitative analysis agent is used to determine the weights of risk factors using the entropy weight method and to calculate the posterior probability of the path using a Bayesian network, thereby quantifying the probability and uncertainty of risk occurrence.

4. The method for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling as described in claim 3, characterized in that, Based on the extracted key information and the unified dataset, a Transformer architecture finely tuned for the deep-sea drilling field is used to construct and dynamically update a three-level blowout risk knowledge graph for the deep-sea high-temperature and high-pressure drilling to be tested, including: Using a semantic understanding agent, based on the Transformer architecture fine-tuned in the field of deep-sea drilling and the extracted key information, a large model is pre-trained. Risk source entities, influencing factor entities, accident consequence entities, and the relationships between various entities are extracted from the unified dataset, and a fine-tuned large model is obtained. Using the ISM algorithm, based on the extracted key information and the extracted risk source entities, influencing factor entities, accident consequence entities, and the relationships between various entities, a three-level blowout risk knowledge graph is constructed, including a risk source layer, a transmission path layer, and an accident consequence layer. The three-level blowout risk knowledge graph is G(V, E, W), where V is the risk node, E is the edge, and W is the edge weight. By acquiring sensor and log data, and employing incremental learning algorithms to update the entity extraction and relationship recognition capabilities of the large model, incremental updates of the large model are achieved, thereby generating a dynamic knowledge graph that includes several types of risk nodes and related relationships.

5. The method for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling as described in claim 4, characterized in that, The three-level blowout risk knowledge graph is a directed hierarchical network. The flow of risk is from bottom to top, that is, from the risk source layer to the transmission path layer, and finally converges to the accident consequence layer. The risk source layer includes the root causes of the risk, the transmission path layer includes the intermediate states and key events in the process of risk evolving from the source to the consequence, and the accident consequence layer includes the eventual undesirable events that may occur.

6. The method for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling as described in claim 3, characterized in that, The method employs a pre-built multi-agent collaborative architecture, traverses a three-level blowout risk knowledge graph, identifies the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling to be tested, and visualizes them, including: A data-aware intelligent agent is used to continuously monitor real-time data streams. When an anomaly is detected, a path reasoning intelligent agent is used to traverse the three-level blowout risk knowledge graph and combine it with the attention mechanism of the large model to identify and filter all potential risk evolution paths, and select the core risk evolution paths that meet the requirements. A quantitative analysis agent was used to determine the weights of each risk factor in the three-level blowout risk knowledge graph and calculate the posterior probability of the core risk evolution path by combining the entropy weight method and Bayesian network, thus obtaining the quantitative analysis results. The ForceAtlas layout algorithm is used to construct a dual visualization mode of "knowledge graph + risk heat map" and the node styles are designed differently according to risk categories to visualize the evolution path of the selected core risks.

7. The method for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling as described in claim 2, characterized in that, The aforementioned pre-built multi-agent collaborative architecture generates a decision support report based on the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling to be tested, including: Based on the heat value of risk nodes and the path correlation strength coefficient, set up multi-level early warning; Using a semantic understanding agent, a decision support report is generated based on visualization results, multi-level early warning settings, the core risk evolution path of the deep-sea high-temperature and high-pressure drilling to be tested, and quantitative analysis results.

8. A system for identifying the evolution path of blowout risk in deep-sea high-temperature and high-pressure drilling, characterized in that, include: The key information extraction module is used to extract key information from historical blowout accident reports, safety procedures, and expert experience documents of the deep-sea high-temperature and high-pressure drilling wells to be tested. The unified dataset construction module is used to collect real-time operating data, equipment operation data, performance parameters and geological environment data of the deep-sea high-temperature and high-pressure drilling to be tested, using a pre-built multi-agent collaborative architecture, and then preprocess the data to form a unified dataset. The knowledge graph construction module is used to construct and dynamically update a three-level blowout risk knowledge graph for the deep-sea high-temperature and high-pressure drilling to be tested, based on the extracted key information and the unified dataset formed, using a Transformer architecture finely tuned for the deep-sea drilling field. The core risk evolution path identification module is used to identify the core risk evolution path and quantitative analysis results of the deep-sea high-temperature and high-pressure drilling to be tested by traversing the three-level blowout risk knowledge graph using a pre-built multi-agent collaborative architecture, and then visualize them.

9. A processing device, characterized in that, The system includes a computer program, wherein when the computer program is executed by a processing device, it is used to implement the steps corresponding to the deep-sea high-temperature and high-pressure drilling blowout risk evolution path identification method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, it is used to implement the steps corresponding to the deep-sea high-temperature and high-pressure drilling blowout risk evolution path identification method according to any one of claims 1-7.

Images