A Data-Driven Virtual Equipment Modeling Method for Fault Diagnosis of Large Rotating Equipment

By constructing a data-driven virtual equipment model, the problem of abstract classification and modeling of large rotating equipment in refining and chemical plants was solved, data association of various types of equipment was realized, the comprehensiveness of fault diagnosis and data support were improved, and a detailed fault history database was formed.

CN117216922BActive Publication Date: 2026-11-10PETROCHINA CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202210611713.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-05-31
Publication Date
2026-11-10
Estimated Expiration
2042-05-31

AI Technical Summary

Technical Problem

Existing technologies lack methods for abstracting, classifying, modeling, and associating data of various types of large rotating equipment in refining and chemical plants for fault diagnosis. This makes it difficult to provide a unified data model, resulting in an incomplete and uncommon fault diagnosis system in the refining and chemical industry.

Method used

By using a data-driven approach, a set of structural objects, a measurement/sampling set, a structural hierarchy object model, and a mapping of physical equipment are established to construct a virtual equipment model. This enables the abstract expression and data collection of large rotating equipment in refining and chemical plants, thus forming a virtual equipment model.

Benefits of technology

It has improved the comprehensiveness and universality of fault diagnosis for large rotating equipment in refining and chemical plants, provided a good data foundation for fault diagnosis, supported health assessment and fault prediction, and formed a large database of equipment operating status and fault history.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117216922B_ABST
    Figure CN117216922B_ABST
Patent Text Reader

Abstract

This invention discloses a data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment. It analyzes discrete physical equipment objects and their measuring points, achieving hierarchical abstraction and classification to construct a structural object model. A mapping is then performed between the physical equipment's basic data, measuring points, and the structural object model to establish a virtual equipment model. This data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment provides a solid diagnostic data foundation for data-driven fault diagnosis by mapping physical equipment data to establish a virtual equipment model. The structural object model aggregates data, providing more data for training and validating the data-driven fault diagnosis model, thus improving the comprehensiveness and universality of fault diagnosis. This method has positive guiding significance for research on fault diagnosis of large rotating equipment in refining and chemical plants.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the technical field of fault diagnosis methods for large rotating equipment (units) in refining and chemical plants, specifically involving a data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment. Background Technology

[0002] In recent years, to ensure smooth production, protect personal safety, and prevent sudden equipment failures from impacting the overall safety and stability of the production process, numerous enterprises both domestically and internationally have successfully implemented equipment fault diagnosis, researching and developing various methods and systems for fault diagnosis based on simulation calculations, data fusion, and data mining. For the refining and chemical industry, with its extremely high and stringent requirements for production stability and safety, there is a growing need to enhance predictive maintenance capabilities and increase research and support for condition monitoring, fault diagnosis, and health assessment technologies for large rotating equipment (units). This necessitates the development of health assessment and fault diagnosis systems suitable for large rotating equipment (units) in refining and chemical plants. Due to the diverse types of large rotating equipment (units) in refining and chemical plants, there is an urgent need for an effective equipment object model design and construction method to provide a unified and universal data model support for data-driven health assessment and fault diagnosis systems.

[0003] Currently published patents primarily focus on fault diagnosis methods and systems for specific equipment such as generator sets, wind turbines, and gas turbine generators. No content on establishing diagnostic object models that implements functions such as abstract classification modeling and data association of physical equipment has been found. Furthermore, comprehensive analysis and macroscopic modeling of various types of large rotating equipment (units) such as compressor units and flue gas turbines in refining and chemical plants are lacking. For example, patents such as "Intelligent Fault Diagnosis System and Method for Industrial Equipment Based on Data Fusion" (CN110377001A), "Fault Diagnosis Method Based on Deep Mining and Learning of Fault Data" (CN110263846A), "Intelligent Diagnosis System and Method for Vibration Faults of Rotating Machinery Based on Simulation Calculation" (CN110579368A), and "A Fault Diagnosis Method for Wind Turbine Units Based on Data Analysis" (CN112462736A) all describe fault diagnosis systems and methods, without involving equipment object classification modeling. Summary of the Invention

[0004] The purpose of this invention is to provide a data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment, providing data model support for the realization of data-driven fault diagnosis.

[0005] The technical solution adopted in this invention is: a data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment, comprising the following steps:

[0006] Step 1: Establish a set of structural objects: Collect and classify the structural information of large rotating equipment (units) in the refining and chemical plant, realize the abstract expression of business object systems, equipment and components, form a set of structural objects, and use the structural BOM (Bill of Material) to represent the relationship between the structure and its constituent substructures.

[0007] Step 2: Establish a measurement / sampling set: Collect and process information on physical measurement points and sampling points of large rotating equipment (units) in the refining and chemical plant, and realize the abstract expression of the structure-based attributes, location, and orientation of the measurement points / sampling points to form a structure-based measurement / sampling set;

[0008] Step 3: Construct a hierarchical object model: Based on the set of structural objects and the structural BOM in Step 1, create a hierarchical object model. According to the set of structural objects and the structural BOM, establish the corresponding upper-level structural objects of the structural objects, and drill down to iterate and generate structural objects until the lowest level structure.

[0009] Step 4: Construct a structural measurement / sampling hierarchy object model: Based on the measurement / sampling set established in Step 2 and the structural hierarchy object model established in Step 3, create a structural measurement / sampling hierarchy object model. Traverse each structural hierarchy object and generate a structural measurement / sampling hierarchy object for each measurement and sampling on its corresponding structure.

[0010] Step 5: Mapping the physical equipment to the structural object model: Map the physical large rotating equipment (unit) to the structural hierarchical object model (system, equipment, component) obtained in Step 3, and map the measurement points / sampling points of the physical large rotating equipment (unit) to the structural measurement / sampling hierarchical object model obtained in Step 4, thereby establishing a virtual equipment model of the large rotating equipment (unit).

[0011] The invention is further characterized in that,

[0012] The structural information of large rotating equipment collected in step 1 includes basic information such as type, parameters and equipment composition, as well as data information on condition monitoring, process operation, laboratory testing and historical maintenance.

[0013] The structural object in step 1 is an individual object with a specific functional location and installation location within a specific organizational structure, and which includes its relevant historical information.

[0014] The system in step 1 is a business entity structure object with a hierarchy, including the system itself (subsystem) or several devices installed on it, covering the operating status and maintenance history.

[0015] The device in step 1 is a business entity structure object, installed in the device layer of the system, covering the operating status and maintenance history.

[0016] The component in step 1 is an abstract structural object that describes the properties of several identical physical parts. It is installed in the component layer of the system or device and only covers the replacement history.

[0017] The structural BOM in step 1 includes the substructures that make up the structure, their quantities, and the installation locations of the structure.

[0018] The physical measurement and sampling point information collected in step 2 for large rotating equipment includes data from condition monitoring, process operation, laboratory testing, and historical maintenance. Measurement / sampling points are the primary recorders of the operating status and process operation parameters of large rotating equipment (units). They can also serve as the primary recorders of the physicochemical properties and composition of equipment parts materials or operating media, and are one of the important business objects for fault diagnosis.

[0019] The beneficial effects of this invention are as follows: The virtual equipment modeling method for fault diagnosis of large rotating equipment based on data-driven methods of this invention establishes a virtual equipment model by mapping physical equipment data, which can provide a good diagnostic data foundation for data-driven fault diagnosis; the structured object model collects data, providing more data for training and verification of the data-driven fault diagnosis model, improving the comprehensiveness and universality of fault diagnosis, and has positive guiding significance for the research on fault diagnosis of large rotating equipment in refining and chemical plants. Attached Figure Description

[0020] Figure 1 This is a diagram showing the structural hierarchy of the virtual equipment modeling method for fault diagnosis of large rotating equipment based on data-driven principles, and its measurement and sampling logic steps.

[0021] Figure 2 This is a diagram showing the relationship between entity objects, structural measurement hierarchy objects, and structural measurement objects in the data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment according to the present invention.

[0022] Figure 3 This is a diagram showing the relationship between entity measurement points / sampling points, structural hierarchy measurement / sampling, and structural measurement / sampling objects in the data-driven virtual equipment fault diagnosis method for large rotating equipment of the present invention. Detailed Implementation

[0023] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0024] This invention provides a data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment. It analyzes discrete physical equipment (unit) objects and their measuring points, realizes hierarchical abstraction and classification, and constructs a structural object model; it maps the basic data of physical equipment, measuring points and structural object model to establish a virtual equipment model.

[0025] Through the above methods, the data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment of the present invention performs structural analysis on commonly used large rotating equipment in the refining and chemical industry, applies abstraction, classification and mapping methods to establish a virtual equipment model of large rotating equipment (units), realizes the construction of case library and expert knowledge base for fault diagnosis of large rotating equipment (units) based on data, thereby improving the relevance and universality of case library and expert knowledge base, and providing effective judgment basis and data support for fault diagnosis of large rotating equipment (units) in refining and chemical plants.

[0026] Example 1

[0027] Step 1: Collect and organize basic information such as the type, parameters, and equipment composition of large rotating equipment (units) in the refining and chemical plant; collect and organize detailed data information from control systems such as condition monitoring, process operation, laboratory testing, and historical maintenance.

[0028] Step 2: Based on the structural information of large rotating equipment (units) collected in Step 1, collect and classify it, establish a set of structural objects, complete the abstract expression of business entity objects such as systems, equipment, and components, and establish a structural BOM to express their compositional relationships.

[0029] Step 3: Based on the physical measurement points and sampling points information collected in Step 1, such as status monitoring, process operation, and laboratory testing, process and classify them to establish a structure-based measurement / sampling set, thereby realizing the abstract expression of the attributes, location, and orientation of the measurement points / sampling points.

[0030] Step 4, as follows Figure 1 As shown, based on the structural objects and structural BOM established in step 2, a structural hierarchy object model is created. According to the structural object set and structural BOM composition, the structural hierarchy objects are generated iteratively by drilling down level by level, and the corresponding upper-level structural hierarchy objects are established until the lowest level structure is reached.

[0031] Step 5, as follows Figure 1 As shown, based on the measurement / sampling object established in step 3 and the structural hierarchy object model established in step 4, a structural measurement / sampling hierarchy object model is created. Each structural hierarchy object is traversed, and a structural measurement / sampling hierarchy object is generated for each measurement and sampling on its corresponding structure.

[0032] Step 6, as follows Figure 2 As shown, based on the basic information data such as the type, parameters, and composition of the physical equipment collected in step 1, the corresponding mapping between large physical rotating equipment (units) and structural hierarchical objects (systems, equipment, components) is carried out.

[0033] Step 7, as follows Figure 3 As shown, the physical measurement points and sampling points information of the physical equipment, such as status monitoring, process operation, and laboratory testing, collected in step 1 are organized, and a mapping is performed between the measurement points / sampling points of the physical large rotating equipment (unit) and the structural measurement / sampling hierarchical object model.

[0034] Step 8: The virtual equipment model of a large rotating equipment (unit) is completed.

[0035] Example 2

[0036] In the above embodiment 1, if there are multiple large rotating equipment (units) of the same type, it is only necessary to abstract and classify them based on the general structure and measuring points of this type of large rotating equipment (units), and construct a set of structural object set, structural BOM, measurement / sampling set, structural hierarchical object model and structural measurement / sampling hierarchical object model. In steps 6 and 7, the structural hierarchical objects and structural measurement / sampling hierarchical objects of multiple physical equipment (units) of this type are mapped respectively, and the virtual equipment model of multiple physical equipment (units) of the same type is established.

[0037] Example 3

[0038] In the above embodiment 1, if a set of structural objects, structural BOM, measurement / sampling set, structural hierarchical object model and structural measurement / sampling hierarchical object model of a certain type of large rotating equipment (unit) already exist, then after step 1 is completed, proceed directly to steps 6, 7 and 8 to complete the establishment of a virtual equipment model of a certain type of physical equipment (unit).

[0039] Example 4

[0040] In practical application, from 2019 to 2020, a regional company of China National Petroleum Corporation applied the data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment proposed in this invention. It constructed general structural object models for compressors, fans, turbines, and generators, mapping data of large rotating equipment (units) in catalytic cracking units, continuous reforming units, and power units. Six sets of virtual equipment models were established, including rich gas compressors (units), circulating hydrogen compressors (units), main fans (units), standby fans (units), and extraction condensing turbines (units). This integrated basic equipment data, equipment status data, process operation parameters, and maintenance and fault history data from the underlying condition monitoring system, the Production Operation Real-Time Database (PHD), the Laboratory Information System (LIMS), and other comprehensive equipment management platforms. This provided a model foundation for the implementation of a data-driven health evaluation and fault diagnosis system for large rotating equipment, forming a large database of equipment operating status and fault history, and continuously accumulating a database of operating and fault cases.

[0041] Results Analysis

[0042] In the construction of a health evaluation and fault diagnosis system for large rotating equipment in refining and chemical plants, this invention is applied to establish a virtual equipment model and map physical equipment data. This enables the integration of basic equipment data, equipment status data, process operation parameters, maintenance and fault history data from the underlying equipment monitoring system, the production operation real-time database (PHD), the laboratory information system (LIMS), and other comprehensive equipment management platforms. This forms a large database of equipment operating status and fault history, and continuously accumulates a database of operating and fault cases.

[0043] Subsequently, during model training and validation, historical maintenance records and inspection / maintenance plans of the equipment are imported into the knowledge base via templates to expand the model, forming a case library and rule library with clearly categorized information on equipment parameters, abnormal states, and equipment faults. Real-time measurement data is processed using big data methods and feature extraction to determine the current state of the equipment. When fault characteristics are detected, they are matched against the existing knowledge base, and handling suggestions are provided. Engineers optimize and improve these suggestions based on the actual on-site handling situation. If necessary, big data experts adjust and correct the backend algorithm.

Claims

1. A data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment, characterized in that, Includes the following steps: Step 1: Collect and classify the structural information of large rotating equipment. The collected structural information of large rotating equipment includes basic information such as type, parameters and equipment composition. This enables the abstract expression of business object systems, equipment and components, forming a set of structural objects. A structural object is an individual object with a specific functional position and installation position under a specific organizational structure, which includes its relevant historical information. The relationship between the structure and its substructures is represented by the structural BOM. The structural BOM includes the substructures that make up the structure, their quantity and installation position. Step 2: Collect and process the physical measurement points and sampling points information of large rotating equipment. The collected physical measurement points and sampling points information of large rotating equipment include data information on condition monitoring, process operation, laboratory testing and historical maintenance. This enables the abstract expression of the structure-based attributes, location and orientation of the measurement points / sampling points, forming a structure-based measurement / sampling set. Step 3: Based on the set of structural objects and the structural BOM in Step 1, create a structural hierarchy object model. According to the set of structural objects and the structural BOM, establish the upper-level structural hierarchy objects corresponding to the structural hierarchy objects, and drill down to iterate and generate structural hierarchy objects until the lowest level structure. Step 4: Based on the measurement / sampling set established in Step 2 and the structural hierarchy object model established in Step 3, create a structural measurement / sampling hierarchy object model. Traverse each structural hierarchy object and generate a structural measurement / sampling hierarchy object for each measurement and sampling on its corresponding structure. Step 5: Map the physical large rotating equipment to the structural hierarchy object model obtained in Step 3, and map the physical large rotating equipment's measuring points / sampling points to the structural measurement / sampling hierarchy object model obtained in Step 4, thereby establishing a virtual equipment model of the large rotating equipment.

2. The data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment as described in claim 1, characterized in that, The system in step 1 is a business entity structure object with hierarchy, including the system itself or several devices installed on it, covering operating status and maintenance history.

3. The data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment as described in claim 1, characterized in that, The device in step 1 is a business entity structure object, installed in the device layer of the system, covering operating status and maintenance history.

4. The data-driven virtual equipment modeling method for fault diagnosis of large rotating equipment as described in claim 1, characterized in that, The component in step 1 is an abstract structural object that describes the attributes of several identical physical parts. It is installed in the component layer of the system or device and only covers the replacement history.

Citation Information

Patent Citations

  • Fault diagnosis method based on fault data deep mining and learning

    CN110263846A

  • Industrial equipment intelligent fault diagnosis system and method based on data fusion

    CN110377001A

  • Simulating calculation based intelligent rotary mechanical fault diagnosis system and method

    CN110579368A

  • Wind turbine generator fault diagnosis method based on data analysis

    CN112462736A

  • Subway train performance evaluation system construction method based on digital twinning

    CN112258094A