Digital twin equipment fault prediction method and system based on knowledge graph

By constructing a panoramic knowledge map in the oil field and combining graph databases and machine learning, the problems of low accuracy and insufficient flexibility in the oil field are solved, and high accuracy and flexibility of fault prediction are achieved, supporting intelligence maintain.

CN119047556APending Publication Date: 2024-11-29SHENZHEN JIAYUNTONG ELECTRONICS

Patent Information

Application Number
CN202411179956.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-27
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

Traditional equipment fault prediction methods have low prediction accuracy in oil fields, difficult to deal with complex and changeable environments, cannot make full use of multi-source heterogeneous data, lack knowledge inference ability, and cannot flexibly deal with unknown fault types.

Method used

Using a digital twin equipment fault prediction method based on knowledge graph, the failure prediction model is built to realize real-time equipment status by constructing a panoramic knowledge graph. Monitor and fault prediction, providing fault warning and decision-making support.

Benefits of technology

It improves the accuracy and real-time nature of fault prediction, enhances the knowledge reasoning and decision-making support capabilities, can flexibly respond to new types of faults, and realizes intelligent maintenance and management of oilfield equipment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119047556A_ABST
    Figure CN119047556A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of oil field equipment, in particular to a digital twin equipment fault prediction method and system based on a knowledge graph, and the method comprises the following steps: 1, constructing and managing the knowledge graph; a knowledge graph technology is utilized to integrate and associate multi-source heterogeneous data to form a structured knowledge system, comprehensive modeling is performed on equipment operation data, historical fault data, expert knowledge and maintenance records, and a panoramic knowledge graph of oil field station equipment is constructed. The method improves the accuracy and real-time performance of fault prediction, optimizes the data integration and management, enhances the knowledge reasoning and decision support capability, and achieves the intelligent maintenance and management of oil field station equipment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of oilfield equipment technology, and in particular to a digital twin equipment fault prediction method and system based on a knowledge graph. Background Art

[0002] Predictive maintenance of oilfield equipment is an important means for oilfields to improve operational efficiency and reduce risks. However, traditional equipment failure prediction methods have low prediction accuracy and difficulty in predicting unknown failure types. The main reasons are:

[0003] 1. Traditional equipment failure prediction methods mainly rely on rule models and historical data analysis, which are difficult to handle the complex and changeable oilfield station equipment environment and have low prediction accuracy;

[0004] 2. Existing methods cannot fully utilize the large amount of multi-source heterogeneous data in oilfield stations (including sensor data, operation logs, equipment information, maintenance records, etc.), making it difficult to build a comprehensive fault prediction model;

[0005] 3. Traditional fault prediction methods lack knowledge reasoning capabilities and are unable to extract effective knowledge from complex fault phenomena and apply it to prediction. They also lack flexibility and scalability when dealing with unknown fault types and emerging problems. Summary of the invention

[0006] The purpose of the present invention is to solve the above-mentioned shortcomings in the prior art and to propose a digital twin device fault prediction method and system based on knowledge graph.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] A digital twin device fault prediction method and system based on knowledge graph is designed. The digital twin device fault prediction method based on knowledge graph is as follows:

[0009] Step 1: Knowledge graph construction and management;

[0010] Using knowledge graph technology, we integrate and associate multi-source heterogeneous data to form a structured knowledge system, conduct comprehensive modeling of equipment operation data, historical fault data, expert knowledge and maintenance records, and build a panoramic knowledge graph of oilfield station equipment;

[0011] Step 2: Construction and optimization of fault prediction model;

[0012] Based on the graph data and time series data of the knowledge graph, we combine the graph database, graph computing technology and machine learning algorithm to build an equipment failure prediction model, realize real-time monitoring of equipment status and prediction and early warning of failures, and help operators take maintenance measures in time to reduce downtime and maintenance costs;

[0013] Step 3: Knowledge reasoning and application;

[0014] Through the knowledge graph, we can realize the reasoning analysis of complex fault phenomena, extract potential fault modes and laws, provide explanations and decision support for fault prediction results, and help operators understand the causes of faults and take corresponding measures.

[0015] In detail, the digital twin device fault prediction system based on knowledge graph is as follows:

[0016] S1. Data management service: collect and manage oilfield knowledge data and operation and maintenance data to build a knowledge graph;

[0017] S2, graph database: stores and manages graph data and provides graph algorithm support;

[0018] S3, edge service: provides the collection and storage of various real-time data of oilfield stations and provides equipment status, alarm, and event information to the graph database;

[0019] S4. Algorithm service: Based on the graph data in the graph database and the time series data in the time series database, Neo4j graph embedding algorithm is combined with machine learning to predict oilfield equipment failures and generate maintenance suggestions.

[0020] S5. Oilfield station digital twin system: Provides visualization and remote control of oilfield station equipment status and process flow, and provides predictive maintenance of equipment in combination with algorithm services.

[0021] In detail, the data management service in S1, the main function of the oilfield station data management service is to collect and manage oilfield knowledge data and operation and maintenance data, and after data cleaning, classification, knowledge extraction and other processing, use the extracted knowledge and data to build an oilfield knowledge graph, including equipment information graph, process flow graph, operation status graph, fault diagnosis graph, etc., to achieve visual representation and correlation analysis of knowledge, and provide intelligent knowledge management and decision support.

[0022] In detail, the graph database in S2, as a core component of the knowledge graph, stores and manages data in a graph structure and can efficiently process data with complex relationships.

[0023] In detail, the system in S2 adopts Neo4j, a mainstream graph database. Neo4j has efficient graph data management, query capabilities and rich graph algorithm functions. The oilfield station equipment fault prediction function of this system is mainly based on the combination of Neo4j graph embedding algorithm and machine learning.

[0024] In detail, the main functions of the edge service in S3, the IoT edge data collection service, include real-time data collection from various program controllers, PLCs, and sensors, and data preprocessing, storage, analysis, and processing at the edge to ensure high-quality data transmission and storage. At the same time, it supports multiple communication protocols to achieve device interoperability, remote management and monitoring of edge devices, deployment and operation of lightweight computing tasks, real-time monitoring and alarm. The edge service stores the real-time collected monitoring data in a time series database, and sends the device status, alarm, and event data to a graph database.

[0025] In detail, the algorithm service in S4, based on the graph data in the graph database and the time series data in the time series database, uses the Neo4j graph embedding algorithm to convert the graph structure information into feature vectors, and combines it with machine learning to perform fault prediction of oilfield equipment, and generates maintenance recommendations based on the prediction results and knowledge data such as equipment information, fault causes, fault phenomena and solutions in the knowledge graph.

[0026] In detail, the algorithm services in S4 mainly include functional modules such as graph embedding algorithm, feature extraction, model training, model deployment, prediction, and generation of maintenance suggestions.

[0027] In detail, the oilfield station digital twin system in S5 improves the visualization and intelligence of operation and maintenance management. It is an important carrier and method for realizing digital delivery requirements. It provides high-precision 3D models and visual displays, simulates and displays equipment operating status and process flows, supports remote control and operation, and displays equipment fault diagnosis and predictive maintenance provided by background algorithm services, optimizes production processes, and realizes full life cycle management and intelligent decision support.

[0028] The digital twin device fault prediction method and system based on knowledge graph proposed in the present invention have the following beneficial effects:

[0029] 1. Improve the accuracy and real-time performance of fault prediction, optimize data integration and management, and enhance knowledge reasoning and decision support capabilities, thereby realizing intelligent maintenance and management of oilfield station equipment.

[0030] 2. Through the construction and maintenance of knowledge graphs and combined with machine learning, this system has good prediction accuracy and adaptability, can flexibly respond to new types of faults, continuously update and optimize fault prediction models, and maintain the long-term effectiveness and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is the overall logic diagram of the present invention;

[0032] Figure 2 It is the process flow chart of the present invention;

[0033] Figure 3 The component knowledge graph of the present invention;

[0034] Figure 4 It is a diagram of the causes, phenomena and solutions of the failures of the present invention;

[0035] Figure 5 It is a flow chart for realizing the algorithm of the present invention;

[0036] Figure 6 Schematic diagram of feature extraction of the present invention

[0037] Figure 7 It is a schematic diagram of equipment failure early warning of the present invention. DETAILED DESCRIPTION

[0038] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0039] Reference Figure 1-Figure 7 , a digital twin device fault prediction method and system based on knowledge graph, characterized in that: the digital twin device fault prediction method based on knowledge graph is as follows:

[0040] Step 1: Knowledge graph construction and management;

[0041] Using knowledge graph technology, we integrate and associate multi-source heterogeneous data to form a structured knowledge system, conduct comprehensive modeling of equipment operation data, historical fault data, expert knowledge and maintenance records, and build a panoramic knowledge graph of oilfield station equipment;

[0042] Step 2: Construction and optimization of fault prediction model;

[0043] Based on the graph data and time series data of the knowledge graph, we combine the graph database, graph computing technology and machine learning algorithm to build an equipment failure prediction model, realize real-time monitoring of equipment status and prediction and early warning of failures, and help operators take maintenance measures in time to reduce downtime and maintenance costs;

[0044] Step 3: Knowledge reasoning and application;

[0045] Through the knowledge graph, we can realize the reasoning analysis of complex fault phenomena, extract potential fault modes and laws, provide explanations and decision support for fault prediction results, and help operators understand the causes of faults and take corresponding measures.

[0046] Improve the accuracy and scope of fault prediction of oilfield equipment, enhance knowledge reasoning and application capabilities, and thus realize intelligent maintenance and management of oilfield equipment.

[0047] It should be further explained that the digital twin equipment fault prediction system based on knowledge graph is as follows:

[0048] S1. Data management service: collect and manage oilfield knowledge data and operation and maintenance data to build a knowledge graph;

[0049] S2, graph database: stores and manages graph data and provides graph algorithm support;

[0050] S3, edge service: provides the collection and storage of various real-time data of oilfield stations and provides equipment status, alarm, and event information to the graph database;

[0051] S4. Algorithm service: Based on the graph data in the graph database and the time series data in the time series database, Neo4j graph embedding algorithm is combined with machine learning to predict oilfield equipment failures and generate maintenance suggestions.

[0052] S5. Oilfield station digital twin system: Provides visualization and remote control of oilfield station equipment status and process flow, and provides predictive maintenance of equipment in combination with algorithm services.

[0053] It should be further explained that the data management service in S1, the main function of the oilfield station data management service is to collect and manage oilfield knowledge data and operation and maintenance data, and after data cleaning, classification, knowledge extraction and other processing, use the extracted knowledge and data to build an oilfield knowledge graph, including equipment information graph, process flow graph, operation status graph, fault diagnosis graph, etc., to achieve visual representation and correlation analysis of knowledge, and provide intelligent knowledge management and decision support.

[0054] 1) Knowledge data collection

[0055] Knowledge data mainly includes: asset information, equipment information, equipment components, process flow, operating specifications, environmental information, oilfield industry knowledge and other related knowledge data.

[0056] Asset information: Asset information includes comprehensive information of all assets within the oilfield ground station, covering comprehensive data of various resources such as equipment, buildings, land, etc., such as asset name, number, location, procurement information, financial information, etc.

[0057] Equipment information: Collect basic information of all equipment in the oilfield station, including equipment model, specification parameters, operating status, equipment maintenance manual, material documents, etc.

[0058] Equipment components: refine the equipment component information, record the components of each device and its functions, location and other detailed information.

[0059] Process flow: Collect the process flow information of oilfield stations, such as crude oil gathering and transportation, heating, oil and gas separation, oil-water separation, water blending, hot washing process, etc., and collect the specific production process, operation steps, upstream and downstream relationships and other information of these processes.

[0060] Operation specification: Operation specification is to ensure safe, efficient and environmentally friendly oilfield surface production and management, covering all aspects of operation, maintenance, safety, environmental protection and emergency response. Such as operation specification, maintenance specification, safety specification, environmental protection specification, emergency management specification, quality management specification, data management specification, etc.

[0061] Environmental information: Specific environmental information includes air quality data (such as PM2.5, SO2, NOx), water quality monitoring data (such as pH value, dissolved oxygen, heavy metal content), soil quality data (such as pollutant content, soil moisture), meteorological data (such as temperature, humidity, wind speed, rainfall), and waste gas and wastewater emission monitoring data.

[0062] Oilfield industry knowledge: knowledge related to oil production, oilfield surface engineering, equipment management, environmental protection, etc.

[0063] 2) Operation and maintenance data collection

[0064] Operation and maintenance data mainly include: fault phenomenon, fault cause, fault solution, fault handling record, equipment maintenance record, asset life cycle record, etc.

[0065] Common equipment failure phenomena: record common equipment failure phenomena and their manifestations during operation.

[0066] Fault causes and solutions: Summarize and record the possible causes and corresponding solutions for each fault phenomenon.

[0067] Fault handling records: Detailed documents recording various equipment or system failures that occur during oilfield production and their handling processes.

[0068] Equipment maintenance records: record various types of information involved in the equipment maintenance process, such as basic maintenance information, fault description, maintenance process, maintenance results, subsequent use of the equipment, etc.

[0069] Asset life cycle records: The asset life cycle records of oilfield ground stations cover the entire process from asset procurement, installation, operation, maintenance to scrapping. The main records include basic asset information (such as name, number, model, location, etc.), procurement and installation date, supplier information, operating status, maintenance and repair records, fault and processing records, performance evaluation, upgrade and transformation history, service life, financial information (such as procurement cost, maintenance costs, depreciation, etc.), scrapping and final disposal, etc.

[0070] 3) Data update and maintenance

[0071] Real-time data collection: Obtain equipment operation data, alarm information, and event information through edge services.

[0072] Data update: Regularly update the data in the graph database to ensure the real-time and accuracy of the data, including the update of dynamic data such as equipment fault handling information and maintenance records.

[0073] 4) Knowledge graph construction

[0074] Graph modeling: Using Neo4j graph database technology, we structured the collected data and built a knowledge graph. Each data node represents a specific piece of information, such as equipment, components, fault phenomenon, fault cause, solution, etc. Edges represent the relationship between nodes, such as the relationship between equipment in the process, the flow of oil, water and gas, the composition relationship between equipment and components, the association between fault phenomenon and cause, etc.

[0075] Data integration: Unify and integrate multi-source heterogeneous data to ensure data integrity and consistency, and convert raw data into a format that can be used by the graph database through steps such as data cleaning and conversion.

[0076] Example graph data such as:

[0077] Figure 2 :Process flow chart;

[0078] Figure 3 :Component knowledge graph;

[0079] Figure 4 : Diagram of fault causes, phenomena and solutions.

[0080] It should be further explained that the graph database in S2, as the core component of the knowledge graph, stores and manages data in a graph structure and can efficiently process data with complex relationships.

[0081] It should be further explained that the system in S2 uses Neo4j, a mainstream graph database. Neo4j has efficient graph data management, query capabilities and rich graph algorithm functions. The oilfield station equipment fault prediction function of this system is mainly based on the combination of Neo4j graph embedding algorithm and machine learning.

[0082] It should be further explained that the main functions of the edge service in S3, the IoT edge data collection service, include real-time data collection from various program controllers, PLCs, and sensors, and data preprocessing, storage, analysis, and processing at the edge to ensure high-quality data transmission and storage. At the same time, it supports multiple communication protocols to achieve device interoperability, remote management and monitoring of edge devices, deployment and operation of lightweight computing tasks, real-time monitoring and alarm. The edge service stores the real-time collected monitoring data in a time series database, and sends the device status, alarm, and event data to the graph database.

[0083] It should be further explained that the algorithm service in S4, based on the graph data in the graph database and the time series data in the time series database, uses the Neo4j graph embedding algorithm to convert the graph structure information into feature vectors, and combines it with machine learning to predict the faults of oilfield equipment, and generates maintenance recommendations based on the prediction results and the knowledge data such as equipment information, fault causes, fault phenomena and solutions in the knowledge graph.

[0084] It should be further explained that the algorithm services in S4 mainly include: graph embedding algorithm, feature extraction, model training, model deployment, prediction, generation of maintenance suggestions and other functional modules.

[0085] The main steps to implement the fault prediction function of oilfield equipment are as follows:

[0086] Knowledge graph construction

[0087] Build a knowledge graph for oilfield ground stations based on data management services and graph databases;

[0088] Graph Algorithm Applications

[0089] Using the graph algorithm library provided by Neo4j, such as graph embedding algorithm and community detection, the data in the graph database is calculated and analyzed to mine potential failure modes and associations. Through graph calculation, abnormal modes and failure signs in equipment operation are identified to provide basic data for failure prediction.

[0090] The present invention uses the Node2Vec module in the Graph Data Science (GDS) of the Neo4j database for graph embedding, and converts the graph structure information into a feature vector (gds.node2vec.write) as the feature data required for subsequent model training.

[0091] Node2Vec is an algorithm for graph embedding that captures the structural information of the graph by learning the vector representation (embedding) of the nodes in the graph.

[0092] The core idea is to generate node sequences by simulating the random walk of nodes, regard these sequences as sentences in the text, and use a method similar to Word2Vec to train the vector representation of the nodes.

[0093] The implementation principle of Node2Vec can be divided into the following steps:

[0094] 1. Random Walk:

[0095] Starting from each node, a node sequence is generated by random walk.

[0096] Node2Vec introduces parameters p and q to control the flexibility of the walk, providing a balance between DFS (depth-first search) and BFS (breadth-first search). The random walk strategy can be achieved by adjusting p and q:

[0097] p controls the probability of returning to the previous step. A higher p value makes the walk more inclined not to return to the previous node, encouraging exploration of new nodes.

[0098] q controls the probability of exploring new nodes. A lower q value makes the walk more inclined to explore local nodes, similar to DFS;

[0099] 2. Generate node sequence:

[0100] For each node in the graph, multiple node sequences (similar to sentences in text) are generated through random walks;

[0101] 3. Training using the Skip-Gram model:

[0102] The generated node sequence is input into the Skip-Gram model for training. The Skip-Gram model aims to optimize the vector representation of a node by predicting the context node of a node in the sequence.

[0103] The goal is to bring neighboring nodes (context nodes) closer in the vector space, thereby capturing local and global structural information of the graph;

[0104] Node2Vec formula:

[0105] In Node2Vec, given node u and node v, the transition probability π uv The calculation formula is:

[0106] π uv =α pq (t,v)·w uv

[0107] Where: w uv is the edge weight.

[0108] α pq (t,v) is the normalized transition probability based on p and q, which indicates the probability of transitioning from node t to node v. It is defined as follows:

[0109]

[0110] 3) Feature extraction

[0111] Use Cypher statements to extract fault-related feature data from the graph database, including equipment operating parameters, fault history records, alarm information, etc.

[0112] The data format is preprocessed, missing values ​​are processed, and then the embedded features are expanded into multiple columns and feature engineering is performed. Finally, the classification features are encoded and standardized. The feature extraction process is shown in Figure 6 .

[0113] 4) Model training

[0114] Using the machine learning framework Sk L Earn, the fault prediction model is trained based on the extracted feature data. The main steps are: data set segmentation, model training, model evaluation, and model optimization.

[0115] First, split the dataset into a training dataset and a test dataset. When training the model, use Random Forest Classifier to train the model, and use the test set to evaluate the performance of the model, such as calculating accuracy, recall, F1 score and other indicators. Finally, based on the model performance and feature importance analysis, adjust the parameters of Node2Vec or the parameters of the random forest model to optimize the model.

[0116] 5) Model deployment and prediction

[0117] Make predictions based on real-time data.

[0118] 6) Store prediction results and output maintenance suggestions

[0119] The results are stored in the Neo4j database, and maintenance recommendations are generated based on the prediction results and the causes of failures in the knowledge graph.

[0120] It should be further explained that the digital twin system of the oilfield station in S5 improves the visualization and intelligence of operation and maintenance management. It is an important carrier and method for realizing digital delivery requirements. It provides high-precision 3D models and visual displays, simulates the operating status and process flow of equipment, supports remote control and operation, and displays equipment fault diagnosis and predictive maintenance provided by background algorithm services, optimizes production processes, and realizes full life cycle management and intelligent decision support.

[0121] The main functions of the digital twin system in equipment failure prediction are:

[0122] 1) Real-time monitoring:

[0123] Status monitoring: Based on the data in the time series database and graph database, the operating status of the equipment is displayed and monitored in real time;

[0124] 2) Fault warning:

[0125] Based on the real-time data of the graph database, the fault prediction algorithm model is used for real-time analysis and calculation to predict possible equipment failures. When it is predicted that the equipment may fail, the system automatically generates fault warning information and notifies relevant maintenance personnel to take timely measures to prevent the failure from expanding.

[0126] 3) Warning explanation:

[0127] Based on the knowledge graph data, provide detailed explanations for the generated fault warning information, including possible causes and solutions of the fault, to help maintenance personnel make decisions;

[0128] 4)Maintenance suggestions:

[0129] Maintenance suggestions generated based on equipment failure prediction algorithm models provide scientific suggestions for equipment maintenance and upkeep, optimize maintenance strategies, and help maintenance personnel make decisions. For equipment failure warning examples, see Figure 7 .

[0130] Through the above technical scheme, the present invention realizes an oilfield station equipment fault prediction and early warning system based on knowledge graph, improves the accuracy and real-time performance of fault prediction, optimizes data integration and management, and enhances knowledge reasoning and decision-making support capabilities, thereby realizing intelligent maintenance and management of oilfield station equipment.

[0131] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A digital twin device fault prediction method and system based on knowledge graph, characterized by: The digital twin device fault prediction method based on knowledge graph is as follows: Step 1: Knowledge graph construction and management; Using knowledge graph technology, we integrate and associate multi-source heterogeneous data to form a structured knowledge system, conduct comprehensive modeling of equipment operation data, historical fault data, expert knowledge and maintenance records, and build a panoramic knowledge graph of oilfield station equipment; Step 2: Construction and optimization of fault prediction model; Based on the graph data and time series data of the knowledge graph, we combine the graph database, graph computing technology and machine learning algorithm to build an equipment failure prediction model, realize real-time monitoring of equipment status and prediction and early warning of failures, and help operators take maintenance measures in time to reduce downtime and maintenance costs; Step 3: Knowledge reasoning and application; Through the knowledge graph, we can realize the reasoning analysis of complex fault phenomena, extract potential fault modes and laws, provide explanations and decision support for fault prediction results, and help operators understand the causes of faults and take corresponding measures.

2. According to the knowledge graph-based digital twin device fault prediction method and system according to claim 1, it is characterized by: The digital twin device fault prediction system based on knowledge graph is as follows: S1. Data management service: collect and manage oilfield knowledge data and operation and maintenance data to build a knowledge graph; S2, graph database: stores and manages graph data and provides graph algorithm support; S3, edge service: provides the collection and storage of various real-time data of oilfield stations and provides equipment status, alarm, and event information to the graph database; S4. Algorithm service: Based on the graph data in the graph database and the time series data in the time series database, Neo4j graph embedding algorithm is combined with machine learning to predict oilfield equipment failures and generate maintenance suggestions. S5. Oilfield station digital twin system: Provides visualization and remote control of oilfield station equipment status and process flow, and provides predictive maintenance of equipment in combination with algorithm services.

3. The digital twin device fault prediction method and system based on knowledge graph according to claim 2 is characterized in that: The data management service in S1, the main function of the oilfield station data management service is to collect and manage oilfield knowledge data and operation and maintenance data, and after data cleaning, classification, knowledge extraction and other processing, use the extracted knowledge and data to build an oilfield knowledge graph, including equipment information graph, process flow graph, operation status graph, fault diagnosis graph, etc., to achieve visual representation and correlation analysis of knowledge, and provide intelligent knowledge management and decision support.

4. The digital twin device fault prediction method and system based on knowledge graph according to claim 2 is characterized in that: The graph database in S2 is a core component of the knowledge graph. It stores and manages data in a graph structure and can efficiently process data with complex relationships.

5. The digital twin device fault prediction method and system based on knowledge graph according to claim 4 is characterized in that: The system in S2 adopts Neo4j, a mainstream graph database. Neo4j has efficient graph data management, query capabilities and rich graph algorithm functions. The oilfield station equipment fault prediction function of this system is mainly based on the combination of Neo4j graph embedding algorithm and machine learning.

6. The digital twin device fault prediction method and system based on knowledge graph according to claim 2 is characterized in that: The edge service in S3, the main functions of the IoT edge data collection service include real-time data collection from various program controllers, PLCs, and sensors, and data preprocessing, storage, analysis, and processing at the edge to ensure high-quality data transmission and storage. At the same time, it supports multiple communication protocols to achieve device interoperability, remote management and monitoring of edge devices, deployment and operation of lightweight computing tasks, real-time monitoring and alarm. The edge service stores the real-time collected monitoring data in the time series database, and sends the device status, alarm, and event data to the graph database.

7. The digital twin device fault prediction method and system based on knowledge graph according to claim 2 is characterized in that: The algorithm service in S4 is based on the graph data in the graph database and the time series data in the time series database, and uses the Neo4j graph embedding algorithm to convert the graph structure information into a feature vector, and combines it with machine learning to perform fault prediction for oilfield equipment, and generates maintenance recommendations based on the prediction results and knowledge data such as equipment information, fault causes, fault phenomena and solutions in the knowledge graph.

8. The method and system for predicting faults of digital twin equipment based on knowledge graph according to claim 7 is characterized in that: The algorithm services in S4 mainly include functional modules such as graph embedding algorithm, feature extraction, model training, model deployment, prediction, and generation of maintenance suggestions.

9. The digital twin device fault prediction method and system based on knowledge graph according to claim 2 is characterized in that: The oilfield station digital twin system in S5 improves the visualization and intelligence of operation and maintenance management. It is an important carrier and method for realizing digital delivery requirements. It provides high-precision 3D models and visual displays, simulates and displays equipment operating status and process flows, supports remote control and operation, and displays equipment fault diagnosis and predictive maintenance provided by background algorithm services, optimizes production processes, and realizes full life cycle management and intelligent decision support.

Citation Information

Patent Citations

  • Fault prediction method and fault prediction device for distributed system

    CN113961424A

  • Wind turbine generator fault prediction method fusing knowledge graph and deep learning

    CN116187508A

  • Heating furnace digital twin edge method, system and device based on knowledge graph

    CN116701653A

  • Digital twinborn construction method of household paper folding equipment based on knowledge graph

    CN117371201A

  • System and methods for automatic medical knowledge curation

    US20210375488A1

Cited By

  • Event digital twinning method, system and equipment

    CN119862768A

  • Digital operation management system and method

    CN120122585A

  • A digital operation management system and method

    CN120122585B

  • Remote fault diagnosis method and system for sand absorbing, screening and separating all-in-one machine based on Internet of Things

    CN120293233A

  • Dynamic risk analysis method and system based on knowledge graph and digital twinborn model

    CN120911974A