A complex equipment whole life cycle information physical fusion method

Abstract

The application discloses a complex equipment whole life cycle information physical fusion method. The method comprises the following steps: firstly, a four-dimensional space expression of complex equipment whole life cycle business activities is established, and the complex equipment whole life cycle business activities are abstracted to four dimensions of time dimension, data dimension, model dimension and structure dimension; secondly, a whole life cycle data system is constructed based on the fusion of the data dimension and the time dimension; thirdly, various models of the whole life cycle are constructed based on the fusion of the model dimension and the time dimension; then, knowledge mining and fusion are carried out for twinborn applications, and knowledge services are formed through knowledge reasoning; finally, multi-dimensional fusion digital twinborn applications are realized based on the structure dimension, the data dimension and the model dimension. The application solves the problem of how to realize multi-dimensional information physical fusion of data and models in the whole life cycle process of complex equipment and carry out twinborn applications.

Description

Technical Field

[0001] This invention relates to the field of cyber-physical fusion throughout the lifecycle of equipment, and more specifically to a method for cyber-physical fusion throughout the entire lifecycle of complex equipment. Background Technology

[0002] The entire lifecycle of complex equipment involves numerous business activities, generating lifecycle data characterized by inconsistencies, heterogeneity, multiple sources, numerous isolated data points, high levels of noise, and difficulties in data collection. Effectively collecting, organizing, integrating, and fusing this data to form a unified and consistent data system, and processing and mining this complex data, can help solve many common problems such as low R&D efficiency, difficulty in controlling manufacturing quality, and difficulty in reducing operation and maintenance costs. With the development of the Internet of Things (IoT) and sensing technologies, real-time data collection throughout the entire lifecycle of complex equipment has become possible. Simultaneously, the correlation between network and physical systems enables the effective integration of equipment network data and physical data, forming a Cyber-Physical System (CPS), providing a foundation for data integrity for data integration and consistent representation.

[0003] However, existing research still lacks an effective framework and methodology to guide the application of cyber-physical fusion throughout the entire lifecycle of complex equipment. Regarding the complete representation of data, due to differences in subjects and data types, managing data at each stage of the complex equipment lifecycle presents significant challenges, necessitating an advanced and effective technological tool to collaboratively utilize data from all stages. In terms of constructing a full-lifecycle digital twin model, the sheer volume and variety of data involved in the complete representation of the complex equipment lifecycle, coupled with a lack of theoretical methods for comprehensively utilizing this data, makes the construction of a full-lifecycle digital twin model difficult. Regarding application models based on full-lifecycle cyber-physical fusion, existing research mostly focuses on a single stage of the equipment lifecycle, neglecting the correlation between multi-source data throughout the lifecycle, making it difficult to construct a closed-loop optimization system. Therefore, how to effectively utilize the complete representation of the complex equipment lifecycle, conduct effective multi-source data correlation knowledge mining and fusion, and ultimately achieve continuous closed-loop iterative optimization of the equipment, has become a pressing issue to be addressed. Summary of the Invention

[0004] To achieve the above-mentioned objectives, the present invention provides the following technical solution:

[0005] A cyber-physical fusion method for the entire lifecycle of complex equipment includes:

[0006] Step S1: Divide the complex equipment into four dimensions and multiple levels of structure;

[0007] Step S2: Constructing a full lifecycle data system based on the fusion of data and time dimensions;

[0008] Step S3: Constructing various types of models throughout the entire lifecycle based on the fusion of model dimension and time dimension;

[0009] Step S4: Knowledge mining, fusion, and services for twin applications;

[0010] Step S5: Digital twin application based on the multi-dimensional fusion of structural dimension, data dimension, and model dimension;

[0011] Step S1 includes: Step S11: Four-dimensional definition of complex equipment; Step S12: Multi-level structure division; Step S13: Establishment of a coding system for complex equipment structure.

[0012] In step S11: the four dimensions of complex equipment are defined as: time dimension Dim 1 T Data Dimension Dim 2 D Model dimension Dim 3 M Decomposition of structural dimensions Dim 4 Layer Specifically, it includes the following steps S111-S114;

[0013] Step S111 defines the time dimension as Dim 1 T ,use Dim 1 T i This refers to different time phases throughout the lifecycle of complex equipment, including the design phase. Dim 1 T 1 Manufacturing stage Dim 1 T 2 and maintenance phase Dim 1 T 3 ;

[0014] Step S112 defines the data dimension as Dim 2 D ;use Dim 2 D i This represents various types of data generated during the design, manufacturing, and operation and maintenance phases of complex equipment; the data from these three phases are processed through a dataset { Dim 2 Di SetData} to represent, i =1,2,3, where { Dim 2 D 1 SetData} represents the dataset for the design phase, { Dim 2 D 2 SetData} represents the manufacturing stage dataset, { Dim 2 D 3 SetData} represents the dataset for the operation and maintenance phase;

[0015] Step S113 defines the model dimension as Dim 3 M ;use Dim 3 M i This represents the various models constructed during the design, manufacturing, and operation and maintenance phases of complex equipment; it also represents the models throughout the entire lifecycle of complex equipment as a model set { Dim 3 M i SetModel} to represent, i =1,2,3, where { Dim 3 M 1 SetModel} represents the design phase model set, { Dim 3 M 2 SetModel} represents the manufacturing stage model set, { Dim 3 M 3 SetModel} represents the set of models for the operation and maintenance phase;

[0016] Step S114 defines the decomposition structure dimension as Dim 3 Layer ,use Dim 3 Layer i This indicates the structural composition of complex equipment; including the overall machine layer. Dim 3 Layer 1 System layerDim 3 Layer 2 Subsystem layer Dim 3 Layer 3 Component layer [[ID=4 3 ​ 4 Component layer ​ 3 ​ 5 ;

[0017] Step S13 establishes a coding system for complex equipment structures, including complex equipment coding, complex equipment structure object coding, complex equipment lifecycle coding, complex equipment attribute parameter coding, and complex equipment lifecycle model coding. Step S13 includes steps S131-S135.

[0018] Step S131 specifies the rules for the composition of complex equipment instance codes; the complex equipment instance code is defined as a code consisting of 5 groups of integers and strings, with the following structure: equipment code-major category-minor category-equipment type-equipment sequence number;

[0019] Step S132 specifies the coding rules for complex equipment structure objects; the coding of complex equipment structure objects is defined as a code consisting of 7 groups of integers and strings, with the following structure: structure code-complete machine-system-subsystem-component-subcomponent-part;

[0020] Step S133 specifies the life cycle coding of complex equipment; the life cycle stage coding of complex equipment is defined as a code consisting of 4 sets of integers and strings, with the following structure: complex equipment life cycle stage code - cycle stage - sub-stage - sub-process.

[0021] Step S134 defines the encoding of complex equipment attribute parameters; the encoding of complex equipment attribute parameters is defined as an encoding consisting of 7 groups of integers and strings, with the following structure: code-category-major category-intermediate category-first-level classification-second-level classification-attribute parameter;

[0022] Step S135 specifies the coding of the complex equipment life cycle model; the coding of the complex equipment life cycle model is defined as a code consisting of 4 sets of integers and strings, with the following structure: complex equipment life cycle model code - model classification - first-level sub-model - second-level sub-model.

[0023] Step S2 includes: Step S21: Full lifecycle information physical data acquisition; Step S22: Source data processing and establishment of meta dataset; Step S23: Vertical fusion of data dimension and structural dimension; Step S24: Horizontal fusion of data dimension and time dimension.

[0024] Step S3 includes: Step S31: Model composition and association at each stage; Step S32: Internal fusion of models at each stage to achieve dynamic updates; Step S33: Mutual fusion of multiple types of models to form twin models; Step S34: Mutual mapping and association of twin models at each stage.

[0025] Step S4 includes: Step S41: Data mining and knowledge extraction; Step S42: Ontology modeling that integrates knowledge services and context; Step S43: Ontology fusion; Step S44: Knowledge reasoning and intelligent services.

[0026] Step S5 includes: Step S51: Fusion of sibling node data in the metastructure tree with the model set; Step S52: Fusion of parent and child node data in the metastructure tree with the model; Step S53: Execution and result output of the digital twin application;

[0027] In step S51, all data and models on the metastructure tree are first merged at the same level node. Data within the same node is merged in step S2, and models within the same node are merged in step S3. Intelligent computing and data mining are then performed in step S4 to form a mutual fusion of data and models within the same node.

[0028] In step S52, the data and models associated with each node in the metastructure tree are combined with the data and models of the parent node and the child node. Data combination includes data fusion between parent and child nodes through step S2, while model combination involves model fusion between parent and child nodes through step S3. Then, intelligent computing and data mining are performed through S4 to form a mutual fusion process of data and models between parent and child nodes.

[0029] In step S53, the equipment data system, twin model, and knowledge are taken as input. The existing historical data is used to form a knowledge model expressed by a knowledge graph through state judgment calculation, fault prediction calculation, correlation analysis, and data mining. Then, real-time data is taken as input, and relevant knowledge models are retrieved from the knowledge graph according to the data characteristics for calculation, and performance degradation analysis and life prediction twin applications are performed.

[0030] In step S12, a meta-structure tree is used as an index for the entire lifecycle process data and model to achieve consistent identification of data and model; the specific implementation process of step S12 includes the following steps S121-S124;

[0031] Step S121 defines the meta-structure tree and the complex equipment decomposition structure dimension in S11; where each intermediate node on the meta-structure tree represents the complex equipment that can be further broken down into various subsystems and components; the construction of the minimum node of the meta-structure is the indivisible design component, the minimum assembly and manufacturing unit, and the minimum maintenance unit.

[0032] Step S122 further divides the meta-structure tree into the design meta-structure tree, the manufacturing meta-structure tree, and the operation and maintenance meta-structure tree;

[0033] Step S123 maps the metastructure tree to the bill of materials for the design, manufacturing, and operation and maintenance phases using a coding system;

[0034] Step S124 divides the complex equipment into 6 levels from top to bottom, namely the complete machine level. ​ 4 ​ 1. System layer ​ 4 ​ 2 Subsystem layer ​ 4 ​ 3. Component layer ​ 4 ​ 4. Sub-component layer ​ 4 ​ 5. Component layer ​ 4 ​ 6.

[0035] Step S21 includes steps S211-S217;

[0036] Step S211 defines the data sources for the entire lifecycle of complex equipment, including methods such as manual form filling, on-site data collection, and system generation;

[0037] Step S212 defines the data types to be collected;

[0038] Step S213 defines the data format to be collected;

[0039] Step S214 defines the data format and corresponding acquisition method;

[0040] Step S215 involves organizing the complex equipment's full lifecycle information physical data according to the data dimension composition format defined in step S112, and then combining the design phase dataset { ​ 2 D 1 ​} Divide into design requirement dataset { ​ 2D 1 ​ Design a geometric dataset { ​ 2 D 1 ​ Design a physical dataset { ​ 2 D 1 ​}、Design process dataset{ ​ 2 D 1 ​ Design and analyze datasets { ​ 2 D 1 ​} and design test dataset { ​ 2 D 1 ​};

[0041] Step S216 involves composing the complex equipment's full lifecycle information physical data according to the data dimension format defined in step S112, and then combining the manufacturing stage dataset { ​ 2 D 2 ​} Divided into manufacturing planning dataset { ​ 2 D 2 ​}, Manufacturing resource dataset { ​ 2 D 2 ​ Manufacturing execution dataset { ​ 2 D 2 ​ Manufacturing quality dataset { ​ 2 D 1 ​ Manufacturing cost dataset { ​ 2 D 2 ​};

[0042] Step S217 will process the dataset for the operation and maintenance phase { ​ 2D 3 ​} is divided into operation and maintenance status datasets { ​ 2 D 3 ​}、Operations and Maintenance Fault Dataset{ ​ 2 D 3 ​}、Operation and maintenance historical dataset{ ​ 2 D 3 ​}、Operation and maintenance cost dataset{ ​ 2 D 3 ​}、Operations and Maintenance Plan Dataset{ ​ 2 D 3 ​}、 Operation and maintenance record dataset{ ​ 2 D 3 ​}

[0043] Step S22 includes steps S221-S223;

[0044] In step S221, a meta-dataset is formed. The meta-dataset is a collection of data with a certain purpose or similarity formed by clustering different datasets or subsets.

[0045] Step S222 analyzes the data format and processes the source data through a one-sided algorithm processing set to create standardized static and dynamic meta-datasets from the physical and information spaces.

[0046] Step S223 fuses the metadata dataset by using a set of fusion algorithms to collect, process anomalies, and transform data to create a complete set of complex equipment metadata datasets.

[0047] Step S23 associates the metadata datasets formed in each cycle of complex equipment design, manufacturing, and operation and maintenance with the hierarchical structural points of the equipment structure. The data association algorithm used is the association algorithm in the fusion algorithm set. With the metadata structure tree as the vertical axis, a vertical organization of the datasets with the metadata structure tree as the core is formed within each cycle stage. In the design stage, data vertical fusion is carried out based on the product prototype and the metadata structure tree organization. In the manufacturing stage, data vertical fusion is carried out based on product process planning and production organization. In the operation and maintenance stage, data vertical fusion is carried out based on product status monitoring, maintenance plan, and maintenance process organization.

[0048] Step S24 is based on the coding system built on the metastructure tree after the digital prototype is formed. The mapping algorithm in the fusion algorithm set maps the design stage dataset to the next two cycle stages, forming a horizontal data system with time dimension as the core required by each stage, realizing horizontal data fusion. Horizontal data fusion is carried out from the design stage to the manufacturing stage, and the design data is transferred according to the product model data attached to each node of the metastructure tree. From the manufacturing stage to the operation and maintenance stage, the manufacturing data is transferred according to the manufacturing history attached to each node of the metastructure tree. From the design stage to the operation and maintenance stage, the design and manufacturing data are transferred according to the design and manufacturing history attached to each node of the metastructure tree.

[0049] Step S31 includes steps S311-S317;

[0050] Step S311 defines the composition of the design phase model set, which is the design information model set { in the design phase. ​ 3 M 1 ​ Includes prototype models, requirement models, product models, and process models; design mechanism model set { ​ 3 M 1 ​ Includes motion models, dynamic models, constitutive models, and analytical models; design domain model set { ​ 3 M 1 ​ Includes whole-machine optimization model, component optimization model, and performance improvement model;

[0051] Step S312 defines the composition of the manufacturing stage model set, in which the manufacturing information model set { ​ 3 M 2 ​ Includes process models, cost models, quality models, resource models; manufacturing mechanism model set { ​ 3M 2 ​ Includes production system operation mechanism models, equipment motion, function, and performance models; manufacturing domain model set { ​ 3 M 2 ​ It includes a production system control model, a quality analysis model, and an equipment operation and maintenance model;

[0052] Step S313 defines the operation and maintenance phase model set { ​ 3 M 3 ​ The composition of}, in the operation and maintenance phase, the operation and maintenance information model set { ​ 3 M 3 ​} Includes product models, maintenance plan models, knowledge models, cost models; and operation and maintenance mechanism model sets { ​ 3 M 3 ​ Includes motion models, dynamic models, constitutive models, and analytical models; operation and maintenance domain model set { ​ 3 M 3 ​ It includes fault prediction models, condition assessment models, life prediction models, and early warning models;

[0053] Step S314 defines the design phase model set { ​ 3 M 1 ​ The relationships between the three models contained within};

[0054] Step S315 defines the manufacturing stage model set { ​ 3 M 2 ​ The relationships between the three models contained within};

[0055] Step S316 defines the operation and maintenance phase model set { ​ 3 M 2 ​ The relationships between the three models contained within};

[0056] Step S317 defines the interaction relationships between the three phase model sets.

[0057] Step S33 includes steps S331-S333;

[0058] Step S331: Based on the complex equipment's structural composition, working principle, operating environment, status, and twin application requirements, form an information model, mechanism model, and domain model encompassing the entire lifecycle. Through model classification, form model sets for each stage and type.

[0059] Step S332, based on the application scenarios and needs of twin models, selects the appropriate models from the model set to construct twin models, and forms twin effects through data association between various types of models;

[0060] Step S333 defines the role and process of the twin model in each stage. The twin role and process include design optimization and improvement in the design stage, production quality and production cost control in the manufacturing stage, and fault diagnosis, performance evaluation and analysis, and life prediction of each node in the operation and maintenance stage.

[0061] Step S34 includes steps S341-S343;

[0062] In step S341, during the application of the twin model, the twin model is updated and optimized based on equipment history data analysis and knowledge graph services. Furthermore, for the complex equipment lifecycle business processes and activities, the interaction relationship between twin models is established through the mapping between twin models at different stages.

[0063] Step S342 transfers the model generated in the design phase to the manufacturing phase through the meta-structure tree, and feeds back to the design phase through the role of the manufacturing process to update various models;

[0064] Step S343 maps the manufacturing stage model to the operation and maintenance stage using a mapping algorithm from the model fusion algorithm set;

[0065] Step S344 maps the design phase model to the operation and maintenance phase to understand the performance, remaining lifespan, etc. of each node on the equipment metastructure tree. The results of the operation and maintenance phase twin model will also be fed back to the design phase, forming the demand points for design improvement and driving the closed-loop feedback of the entire life cycle of the complex equipment.

[0066] Step S41 includes step S41;

[0067] In S411, for the model dimension ​ 3 M We mine and extract various model sets of complex equipment in the design, manufacturing and operation and maintenance stages to obtain the data-driven analysis and processing model class knowledge contained in these model sets.

[0068] In S412, association rule knowledge contained in the data is mined and extracted. This type of knowledge comes from the mining of datasets in various fields of equipment and parts.

[0069] In S413, user intent recognition knowledge is mined and extracted from the data. This user intent recognition knowledge comes from the mining of datasets in various fields of equipment and parts, as well as the extraction of unstructured knowledge. User intent recognition knowledge is used to fuzzily identify user intent and return results according to user intent, or to provide proactive push services to users, including word segmentation processing models and intent recognition models.

[0070] In S414, external knowledge acquired through data collection is mined and extracted. This type of knowledge comes from internet data in the field and is obtained through data collection and semi-structured knowledge extraction techniques.

[0071] In S415, knowledge of relevant application software systems is mined and extracted. This type of knowledge comes from the knowledge accumulated by existing software systems at various stages of complex equipment.

[0072] Step S43 includes steps S431-S432

[0073] Step S431 achieves ontology fusion at the ontology level through concept, relation alignment, and relation reasoning. Based on the similarity calculation of knowledge data attributes corresponding to concepts, it infers the connection between concepts and proposes numerical and textual attribute similarity measurement and overall concept similarity measurement algorithms, thereby achieving ontology-level fusion.

[0074] In step S432, the structured attribute similarity and unstructured attribute similarity are weighted and normalized according to the concept overall similarity measurement algorithm in step S431 to obtain the comprehensive similarity. Then, the optimal candidate sequence is generated, and the entity with the highest similarity and greater than the threshold is selected as the entity that can be aligned.

[0075] Step S433 performs knowledge fusion at the knowledge graph data level; based on step S431, a unified knowledge representation model is established using the network ontology language, knowledge graph embedding is achieved through word vectors, and knowledge is stored using a native graph database. For various knowledge sources and features, structured knowledge is integrated from the database management system through D2R transformation, and a BERT+Bilstm-CRF deep learning model is constructed to obtain unstructured text entities.

[0076] Step S44 includes steps S441-S443;

[0077] Step S441 involves sensing the request context of the user's request and obtaining... ​ 4 ​ i ,​ 1 T , ​ 3 M It then determines the scope of complex product domains, segments user requests into words, calculates the corresponding knowledge domains through word vectors, calls domain intent recognition models to identify and transform user intents, queries the knowledge graph or calls model algorithms to obtain query results and visualize them.

[0078] Step S442 performs sensor encoding acquisition and physical sensing stream data analysis on the physical data-driven automatic request service, identifies the context and knowledge domain and obtains the knowledge domain encoding, and then generates the graph query basis, generates the AI ​​model and generates the function interface based on the knowledge domain encoding. Then, it queries the knowledge graph or calls the model algorithm to obtain the query results and displays them visually.

[0079] Step S443: Construct a knowledge proactive push service, perceive the user's current task, and obtain the meta-model nodes of the equipment structure tree associated with the task. ​ 4 ​ ,Model ​ 3 M and corresponding life cycle stages ​ 1 T The system determines the scope of complex equipment domains, queries the knowledge graph for each knowledge domain and returns the knowledge dataset, generates domain labels, and then dynamically generates push window content.

[0080] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0081] (1) In practice, the inventors discovered that existing technologies lack an abstraction method to address the difficulty of linking and integrating the structure, data, models, and various stages of the lifecycle of complex equipment. To address this technical problem, this invention proposes a four-dimensional spatial representation method for the entire lifecycle business activities of complex equipment. It proposes the concept of a four-dimensional space including time dimension, model dimension, data dimension, and decomposition structure dimension. The expression of the decomposition structure dimension is realized through a metastructure tree coding system. With the decomposition structure dimension as the core, the unified association of the other three dimensions of each node is realized, thereby organizing and managing various information physical data, models, and dynamic activities throughout the entire lifecycle of complex equipment in an orderly manner.

[0082] (2) In practice, the inventors discovered that, under the organization of the time dimension, data from each stage and structural nodes different from those of complex equipment together constitute a series of complex equipment lifecycle data. Based on this, a method for constructing a lifecycle data system based on the fusion of the data dimension and the time dimension was proposed. This method involves four steps: lifecycle data acquisition, source data processing, vertical fusion of the data dimension and the structural dimension, and horizontal fusion of the data dimension and the time dimension. These steps form a complete lifecycle data system for complex equipment. Furthermore, a method for horizontal and vertical data fusion was proposed. Vertical data fusion enables the organization of data within each stage, while horizontal data fusion enables the transfer of data between stages.

[0083] (3) In practice, the inventors discovered that complex equipment contains multiple different types of models at each stage of its life cycle, and that models of the same type and different types interact with each other. To address this, the present invention proposes a method for fusing complex equipment life cycle models based on model dimension and time dimension. By defining information model, domain model, mechanism model, and twin model, the models are classified, and the interaction relationships between various types of models are analyzed, thus realizing the fusion of different models throughout the life cycle of complex equipment.

[0084] (4) In practice, the inventors discovered that in the data system of the entire life cycle of complex equipment, data at each stage and in each dimension are interconnected and influence each other, continuously accumulating to form experience and knowledge. Therefore, this invention proposes a knowledge mining, fusion, and service method for twin applications. It outputs knowledge through data mining, knowledge extraction, ontology modeling, and ontology fusion, and then provides knowledge services for various types of twin applications through knowledge reasoning. This process indicates how to combine data, models, and metastructure trees through knowledge graphs in four-dimensional space to achieve twin applications.

[0085] (5) In practice, the inventors discovered that the fusion of structural dimension, time dimension, model dimension and data of complex equipment provides an effective means for digital twin applications. To this end, this invention proposes a digital twin application based on the multi-dimensional fusion of data dimension, structural dimension and model dimension, and defines the input, output and algorithm set application paradigm of complex equipment digital twin application in four-dimensional space, providing a new approach to solving problems such as performance degradation analysis and fault diagnosis at each stage of the entire life cycle of complex equipment. Attached Figure Description

[0086] ​ A schematic diagram representing the four-dimensional space of complex equipment;

[0087] ​ A schematic diagram of a complex equipment multi-level structure.

[0088] ​ A schematic diagram illustrating the structure and lifecycle attribute coding of complex equipment;

[0089] ​ A schematic diagram illustrating the data fusion throughout the entire lifecycle of complex equipment;

[0090] ​ A schematic diagram illustrating the fusion of a full lifecycle model for complex equipment;

[0091] ​ A schematic diagram of knowledge services based on data mining and knowledge fusion;

[0092] ​ This is a schematic diagram of a digital twin application based on the fusion of structure, data, and model. Detailed Implementation

[0093] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0094] Therefore, the following detailed description of embodiments of the present invention is not intended to limit the scope of the claimed invention, but merely illustrates some embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0095] It should be noted that, unless otherwise specified, the embodiments and features and technical solutions in the present invention can be combined with each other.

[0096] A cyber-physical fusion method for the entire lifecycle of complex equipment includes:

[0097] Step S1: Divide complex equipment into four dimensions and multi-level structures; Step S2: Construct a full lifecycle data system based on the fusion of data and time dimensions; Step S3: Construct various types of full lifecycle models based on the fusion of model and time dimensions; Step S4: Knowledge mining, fusion, and services for twin applications; Step S5: Digital twin applications based on the multi-dimensional fusion of structural, data, and model dimensions.

[0098] Preferably, step S1 includes: step S11: four-dimensional definition of complex equipment; step S12: multi-level structure division; step S13: structure coding.

[0099] Preferably, the implementation of step S11 is as follows: the four dimensions of the complex equipment are defined as: time dimension ( ​ 1 T ), data dimension ​ 2D ), model dimension ( ​ 3 M ), decompose structural dimensions ( ​ 4 ​ Specifically, it includes the following steps S111-S114.

[0100] Preferably, step S111 defines the time dimension as... ​ 1 T ,use ​ 1 T i This refers to different time phases throughout the lifecycle of complex equipment, including the design phase. ​ 1 T 1 Manufacturing stage ​ 1 T 2 and maintenance phase ​ 1 T 3 .

[0101] Preferably, step S112 defines the data dimension as follows: ​ 2 D .use ​ 2 D i This represents various types of data generated during the design, manufacturing, and operation / maintenance phases of complex equipment. The data from these three phases is processed through a dataset { ​ 2 D i ​} to represent, i =1,2,3, where { ​ 2 D 1 ​} represents the dataset for the design phase, { ​ 2 D 2 ​} represents the manufacturing stage dataset, { ​ 2 D 3 ​} represents the dataset for the operation and maintenance phase.

[0102] Preferably, depending on different view requirements, the data dimensions divided in step S112 can be further divided to form more data subsets.

[0103] Preferably, step S113 defines the model dimension as follows: ​ 3 M .use ​ 3 M i This represents the various models constructed during the design, manufacturing, and operation and maintenance phases of complex equipment. These models throughout the entire lifecycle of complex equipment are grouped into a model set { ​ 3 M i ​} to represent, i =1,2,3, where { ​ 3 M 1 ​} represents the design phase model set, { ​ 3 M 2 ​} represents the manufacturing stage model set, { ​ 3 M 3 ​} represents the set of models for the operation and maintenance phase.

[0104] Preferred model sets for each stage { ​ 3 M i ​ Based on model type, models can be further divided into three categories: those based on { ​ 3 M i ​ The information model is represented by ​ 3 M i ​ The mechanism model and the representation of { ​ 3 M i ​ The domain model is represented by}. The fusion of these three models constitutes the digital twin model; the fusion process is described as follows:

[0105] { ​ 3M 1 ​}={{ ​ 3 M i SetInfoModel}∩{Dim 3 M i SetPrinModel}∩ { Dim 3 M i SetFieldModel }}.

[0106] Preferably, step S114 defines the decomposition structure dimension as follows: Dim 3 Layer ,use Dim 3 Layer i This indicates the structural composition of complex equipment. It includes the overall machine layer (…). Dim 3 Layer 1 ), system layer ( Dim 3 Layer 2 ), subsystem layer ( Dim 3 Layer 3 ), component layer ( Dim 3 Layer 4 ), component layer ( Dim 3 Layer 5 (e.g., ..., until an indivisible physical entity.)

[0107] Preferably, the interaction condition between structures at the same level and structures at different levels is physical connection, through which force, motion and information are transmitted between different nodes.

[0108] Preferably, in step S12, a meta-structure tree is used as an index for the entire lifecycle process data and models to achieve consistent identification of data and models. The specific implementation process includes the following steps S121-S124.

[0109] Preferably, step S121 defines the meta-structure tree and the complex equipment decomposition structure dimension in S11. The complex equipment decomposition structure dimension is the decomposition of complex equipment according to its own structure and functional modules.

[0110] Preferably, the metastructure tree includes intermediate nodes and minimum nodes. Each intermediate node in the metastructure tree represents a complex piece of equipment that can be further broken down into subsystems and components. For example, the bogie wheelset component can be divided into two lower-level nodes: wheels and axles. The construction of the minimum node of the metastructure is based on indivisible design components, the smallest assembly and manufacturing unit, and the smallest maintenance unit, such as bearings and springs.

[0111] Preferably, step S122 further divides the metastructure tree into a design metastructure tree, a manufacturing metastructure tree, and an operation and maintenance metastructure tree, with each node of each metastructure tree carrying relevant data, models, and digital twin applications.

[0112] Preferably, in step S123, the metastructure tree is mapped to the bill of materials for the design, manufacturing, and operation and maintenance phases through a coding system, and the composition structure of the complex equipment is determined according to the node division of the bill of materials for the design, manufacturing, and operation and maintenance phases.

[0113] Preferably, step S124 defines the hierarchical division of the complex equipment structure, dividing the complex equipment into 6 levels from top to bottom, namely the whole machine layer. Dim 4 Layer 1. System layer Dim 4 Layer 2 Subsystem layer Dim 4 Layer 3. Component layer Dim 4 Layer 4. Sub-component layer Dim 4 Layer 5. Component layer Dim 4 Layer 6.

[0114] Preferably, step S13 establishes a coding system for complex equipment structures. This coding system implements a naming mechanism for data, models, and files throughout the entire lifecycle of complex equipment, including complex equipment coding, complex equipment structure object coding, complex equipment lifecycle coding, complex equipment attribute parameter coding, and complex equipment lifecycle model coding. The specific implementation process includes the following steps S131-S135.

[0115] Preferably, step S131 specifies the rules for the composition of complex equipment instance codes. The complex equipment instance code is defined as a code consisting of 5 groups of integers and strings, with the structure being "Equipment Code (P) - Major Category (x) - Minor Category (x) - Equipment Class (xx) - Equipment Sequence Number (xxx)".

[0116] Preferably, step S132 specifies the coding rules for complex equipment structure objects. The coding of complex equipment structure objects is defined as a code consisting of 7 groups of integers and strings, with the structure being "Structure Code (Obj) - Whole Machine (x) - System (x) - Subsystem (xx) - Component (xx) - Sub-component (xx) - Part (xxx)".

[0117] Preferably, step S133 specifies the life cycle coding of complex equipment. The life cycle stage coding of complex equipment is defined as a code consisting of 4 sets of integers and strings, with the structure being "complex equipment life cycle stage code (Lcs) - cycle stage (xx) - sub-stage (xx) - sub-process (xx)".

[0118] Preferably, step S134 defines the encoding of complex equipment attribute parameters. The encoding of complex equipment attribute parameters is defined as an encoding consisting of 7 sets of integers and strings, with the structure "code (ObjA)-category (x)-major category (x)-intermediate category (xx)-first-level classification (xx)-second-level classification (xx)-attribute parameter (xxx)".

[0119] Preferably, step S135 specifies the coding of the complex equipment life cycle model. The coding of the complex equipment life cycle model is defined as a code consisting of 4 sets of integers and strings, with the structure being "complex equipment life cycle model code (ObjM) - model classification (xx) - first-level sub-model (xx) - second-level sub-model (xx)".

[0120] In step S1, the four-dimensional spatial representation method for the entire lifecycle business activities of complex equipment proposed in this invention has the following advantages compared with existing technologies: It proposes the concept of a four-dimensional space including time dimension, model dimension, data dimension, and decomposition structure dimension; it realizes the expression of the decomposition structure dimension through a metastructure tree coding system; and it uses the decomposition structure dimension as the core to achieve unified association of the other three dimensions of each node, thereby orderly organizing and managing various cyber-physical data, models, and dynamic activities throughout the entire lifecycle of complex equipment. Furthermore, by specifying the coding rules for complex equipment structural objects, it achieves a unified and standardized naming mechanism for data, models, and files in the lifecycle of complex equipment.

[0121] Preferably, step S2 includes: step S21: full lifecycle information physical data acquisition; step S22: source data processing, establishing a meta dataset; step S23: vertical fusion of data dimension and structural dimension; step S24: horizontal fusion of data dimension and time dimension.

[0122] Preferably, step S21 clarifies the characteristics and acquisition methods of the cyber-physical data of complex equipment throughout its entire life cycle, and the specific implementation process includes steps S211-S217.

[0123] Preferably, step S211 defines the data sources for the entire lifecycle of complex equipment, including methods such as manual form filling, on-site collection, and system generation.

[0124] Preferably, step S212 defines the data types to be collected, including structured and unstructured data such as numerical values, ranges, text, images, videos, and models.

[0125] Preferably, step S213 defines the data formats to be collected, including .doc, .xls, .pdf, .txt, .dwg, .jpg, .mp4, etc.

[0126] Preferably, step S214 defines the data format and corresponding acquisition method. The data format includes streaming data and non-streaming data. Streaming data acquisition involves real-time acquisition at a certain time frequency, such as bearing temperature monitoring data during operation and maintenance, which is collected at regular intervals and continuously accumulated. The streaming data acquisition process places higher demands on data access, storage, and management. Non-streaming data acquisition, on the other hand, involves one-time acquisition, such as part design geometry data. Since non-streaming data may have multiple different versions, version management of the data is necessary.

[0127] Preferably, in step S215, the complex equipment's full lifecycle information physical data is organized according to the data dimension format defined in step S112, and the design phase dataset { Dim 2 D 1 SetData } Divide into design requirement dataset { Dim 2 D 1 SetDataRequirement Design a geometric dataset { Dim 2 D 1 SetDataGeom Design a physical dataset { Dim 2 D 1 SetDataPhysics }、Design process dataset{ Dim 2 D 1 SetDataProcess Design and analyze datasets { Dim 2D 1 SetDataAnalyse } and design test dataset { Dim 2 D 1 SetDataTest }

[0128] Preferably, in step S216, the complex equipment's full lifecycle information physical data is organized according to the data dimension format defined in step S112, and the manufacturing stage dataset { Dim 2 D 2 SetData } Divided into manufacturing planning dataset { Dim 2 D 2 SetDataPlan }, Manufacturing resource dataset { Dim 2 D 2 SetDataResource Manufacturing execution dataset { Dim 2 D 2 SetDataProc-exe Manufacturing quality dataset { Dim 2 D 1 SetDataQuality Manufacturing cost dataset { Dim 2 D 2 SetDataCost }

[0129] Preferably, step S217 will process the dataset from the operation and maintenance phase { Dim 2 D 3 SetData } is divided into operation and maintenance status datasets { Dim 2 D 3 SetDataStatus }、Operations and Maintenance Fault Dataset{ Dim 2 D 3 SetDataFault }、Operation and maintenance historical dataset{ Dim 2 D 3 SetDataRepairHistory }、Operation and maintenance cost dataset{ Dim 2 D 3 SetDataRepairCost }、Operations and Maintenance Plan Dataset{ Dim 2 D 3 SetRepairPlan }、 Operation and maintenance record dataset{ Dim 2 D 3 SetRepairRecord }

[0130] Preferably, step S22 processes the source data, and the specific implementation process includes steps S221-S223.

[0131] Preferably, step S221 defines a meta-dataset, which is a data set with a certain purpose or similarity formed by clustering different datasets or subsets.

[0132] Preferably, step S222 analyzes the data format and processes the source data using a one-sided algorithm. AlgorithmSet_Act_Data To process and create standardized static and dynamic metadata datasets from physical and information spaces. Sub{Sub{Sub{Dim 2 D i SetData}}} It should be noted that this algorithm set is a collection of one-sided data processing algorithms such as acquisition, semantic recognition, cleaning, extraction, and storage, and this algorithm set is existing technology.

[0133] Preferably, step S223 involves fusing the metadata dataset using a set of fusion algorithms. AlgorithmSet_ InfoFusion Data collection, anomaly handling, and transformation are performed to create a complete metadata dataset for complex equipment, achieving physical information fusion. It should be noted that this algorithm set is a collection of data fusion processing algorithms such as classification, association, integration, and mapping. These algorithms process two types of data in two ways, and this algorithm set represents existing technology.

[0134] Preferably, in step S23, the metadata dataset (formed by the classification algorithm in the fusion algorithm set) formed in each cycle of complex equipment design, manufacturing, and operation and maintenance is associated with the hierarchical structural points of the equipment structure (the encoded metadata structure tree). The data association algorithm used is the association algorithm in the fusion algorithm set. With the metadata structure tree as the vertical axis, a vertical organization form of the dataset with the metadata structure tree as the core is formed within each cycle stage, so as to realize the vertical fusion of data.

[0135] Preferably, in step S23, data vertical integration is carried out based on product prototypes and metastructure structure organization during the design phase, based on product process planning and production organization during the manufacturing phase, and based on product status monitoring, maintenance plans, and maintenance process organization during the operation and maintenance phase.

[0136] Preferably, step S24 is based on the coding system built on the metastructure tree after the digital prototype is formed. The design stage dataset is mapped to the next two cycle stages through the mapping algorithm in the fusion algorithm set, thereby forming the horizontal data system with time dimension as the core required by each stage, and realizing horizontal data fusion.

[0137] Preferably, step S24 defines the content and method of data transfer between the three stages of a complex product. From the design stage to the manufacturing stage, design data is transferred based on the product model data attached to each node of the metastructure tree. From the manufacturing stage to the operation and maintenance stage, manufacturing data is transferred based on the manufacturing history attached to each node of the metastructure tree. From the design stage to the operation and maintenance stage, design and manufacturing data are transferred based on the design and manufacturing history attached to each node of the metastructure tree.

[0138] In step S2, the proposed method for constructing a full lifecycle data system based on the fusion of data and time dimensions forms a complete full lifecycle data system for complex equipment through four steps: full lifecycle data acquisition, source data processing, vertical fusion of data and structural dimensions, and horizontal fusion of data and time dimensions. It also proposes a method for horizontal and vertical data fusion, which realizes the organization of data within each stage through vertical data fusion and the transfer of data between each stage through horizontal data fusion.

[0139] Preferably, step S3 includes: step S31: model composition and association at each stage; step S32: model integration at each stage to achieve dynamic updates; step S33: integration of multiple types of models to form twin models; step S34: mutual mapping and association of twin models at each stage.

[0140] Preferably, step S31 further defines the composition and relationship of each type of model at each stage based on step S113, including steps S311-S317;

[0141] Preferably, step S311 defines the composition of the design phase model set, where the design information model set { Dim 3 M 1 SetInfoModel} Includes prototype models, requirement models, product models, and process models; design mechanism model set { Dim 3 M 1 SetPrinModel} Includes motion models, dynamic models, constitutive models, and analytical models; design domain model set { Dim 3 M 1 SetFeildModel} It includes whole-machine optimization models, component optimization models, performance improvement models, etc.

[0142] Preferably, step S312 defines the composition of the manufacturing stage model set, in which the manufacturing information model set { Dim 3 M 2 SetInfoModel} Includes process models, cost models, quality models, resource models; manufacturing mechanism model set { Dim 3 M 2 SetPrinModel} Includes production system operation mechanism models, equipment motion, function, and performance models; manufacturing domain model set { Dim 3 M 2 SetFeildModel} It includes production system management and control models, quality analysis models, and equipment operation and maintenance models.

[0143] Preferably, step S313 defines the operation and maintenance phase model set { Dim 3 M 3 SetModel The composition of}, in the operation and maintenance phase, the operation and maintenance information model set { Dim 3 M 3 SetInfoModel } Includes product models, maintenance plan models, knowledge models, cost models; and operation and maintenance mechanism model sets { Dim 3 M 3 SetPrinModel} Includes motion models, dynamic models, constitutive models, and analytical models; operation and maintenance domain model set { Dim 3 M 3 SetFeildModel} It includes fault prediction models, condition assessment models, life prediction models, and early warning models.

[0144] Preferably, step S314 defines the design phase model set { Dim 3 M 1 SetModel The relationships between the three models contained within}

[0145] Preferably, the design information model set { Dim 3 M 1 SetInfoModel} The included design phase information is input into the design mechanism phase model { Dim 3 M 1 SetPrinModel} The mechanistic model in the design phase reveals the underlying mechanisms, and this design phase information (such as design requirements and product prototypes) is then used to support the domain model set in the design phase. Dim 3 M 1 SetFeildModel} The domain model in the design phase (such as the performance improvement model, component design optimization model, etc.) enables the twin application of the design phase.

[0146] Preferably, step S315 defines the manufacturing stage model set { Dim 3 M 2 SetModel The relationships between the three models contained within}

[0147] Preferably, the manufacturing information model set { Dim 3 M 2 SetInfoModel} The manufacturing stage information included is input into the manufacturing mechanism model set { Dim 3 M 2 SetPrinModel} The mechanistic model in the model reveals the underlying mechanism, and then this manufacturing stage information (such as product process and manufacturing cost) is used to support the manufacturing stage domain model set. Dim 3 M 2 SetFeildModel} The manufacturing stage domain model (such as the production system control model, quality analysis model, etc.) enables the twin application of the manufacturing stage.

[0148] Preferably, step S316 defines the operation and maintenance phase model set { Dim 3 M 2 SetModel The relationships between the three models contained within}

[0149] Preferably, the operation and maintenance information model set { Dim 3 M 3 SetInfoModel} The included operation and maintenance phase information is input into the operation and maintenance mechanism model set { Dim 3 M 3 SetPrinModel} The mechanism model in the middle reveals the mechanism, and then this operation and maintenance phase information (such as maintenance plans and maintenance knowledge) is used to support the operation and maintenance domain model set { Dim 3 M 3 SetFeildModel} The domain models of the operation and maintenance phase (such as fault prediction models, condition assessment models, etc.) are used to realize the twin application of the operation and maintenance phase.

[0150] Preferably, step S317 defines a three-stage model set { Dim 3 M 1 SetModel }、{ Dim 3 M 2 SetModel }and{ Dim 3 M 3 SetModel The interaction between these stages is as follows: the design stage model set maps the model to the manufacturing and operation and maintenance stages, serving as the model foundation for these two stages; the manufacturing stage feeds back design changes that occur during the manufacturing process based on technology and production conditions to the design stage for updating the design stage model; and the operation and maintenance stage feeds back operation and maintenance data to the design stage for optimizing the design scheme.

[0151] Preferably, step S32 takes the information, mechanism, and domain model of each stage as input and employs an algorithm set. {AlgorithmSet_ModelFusion} Through basic model construction, model association, and digital twin operation, various types of digital twin models are output (such as dynamic models in mechanistic models, early warning and prediction models in domain models, and product models in information models). Relationships between these digital twin models are established (e.g., the dynamic model serves as the mechanistic input for the early warning and prediction model, and the product model serves as the information input for the early warning and prediction model), enabling internal fusion and dynamic updates of models at each stage. It should be noted that the algorithm set... {AlgorithmSet_ModelFusion} It is an integration of a series of model fusion algorithms such as classification, association, integration, and mapping. This set of algorithms is an existing technology.

[0152] Preferably, step S33 involves assembling a twin model using multiple types of models and generating twin effects, including twin effects in the design phase, twin effects in the manufacturing phase, and twin effects in the operation and maintenance phase. The specific implementation process is described in steps S331-S333.

[0153] Preferably, step S331 forms an information model, mechanism model, and domain model covering the entire life cycle based on the complex equipment's structural composition, working principle, operating environment, status, and twin application requirements, and forms model sets for each stage and type through model classification.

[0154] Preferably, step S332 is based on the application scenarios and needs of the twin model, selecting the corresponding model from the model set to construct the twin model (such as selecting the bearing dynamics model, fault warning algorithm, and real-time bearing data to construct the bearing fault warning digital twin model for bearing fault warning needs), and forming a twin effect through data association between various types of models.

[0155] Preferably, step S333 defines the role process of the twin model in each stage. The twin role process includes design optimization and improvement in the design stage, production quality and production cost control in the manufacturing stage, and fault diagnosis, performance evaluation and analysis, and life prediction of each node in the operation and maintenance stage.

[0156] Preferably, step S34 defines the process of mapping, association and transmission between the twin models at each stage, and the specific implementation process includes steps S341-S343.

[0157] Preferably, in step S341, during the application of the twin model, the twin model is updated and optimized based on equipment history data analysis and knowledge graph services. Furthermore, for the complex equipment lifecycle business processes and activities, the interaction relationship between twin models is established through mapping between twin models at different stages.

[0158] Preferably, in step S342, the model generated in the design phase is transferred to the manufacturing phase through the meta-structure tree, and the various models are updated by the feedback from the manufacturing process to the design phase.

[0159] Preferably, in step S343, the manufacturing stage model is processed through a model fusion algorithm set. {AlgorithmSet_ ModelFusion} The mapping algorithm in the model is applied to the operation and maintenance phase, and is used for the operation and maintenance of complex equipment to understand the root causes of equipment failures and predict lifespan. The model fusion algorithm set is based on existing technologies.

[0160] Preferably, step S344 maps the design phase model to the operation and maintenance phase to understand the performance, remaining lifespan, etc. of each node on the equipment metastructure tree. The results of the operation and maintenance phase twin model are also fed back to the design phase, forming the demand points for design improvement and driving the closed-loop feedback of the entire life cycle of the complex equipment.

[0161] In step S3, the complex equipment full life cycle model fusion method proposed in this invention, based on model dimension and time dimension, classifies the models by defining information model, domain model, mechanism model and twin model, and analyzes the interaction relationship between various models, thereby realizing the fusion of different models throughout the complex equipment life cycle.

[0162] Preferably, step S4 includes: step S41: data mining and knowledge extraction; step S42: ontology modeling that integrates knowledge services and context; step S43: ontology fusion; step S44: knowledge reasoning and intelligent services.

[0163] Preferably, in step S41, under the twin requirements of each lifecycle stage, knowledge mining, clustering, and extraction are performed on the full lifecycle data and historical business activities to obtain model knowledge, hidden knowledge, and existing knowledge in the system. Then, through the mapping of knowledge to complex equipment structures, the association between nodes of complex equipment structures and knowledge is established. The specific implementation process includes step S41.

[0164] Preferably, in S411, for the model dimension Dim 3 M It includes a set of various models for the design, manufacturing, and operation and maintenance phases of complex equipment. Dim 3 M i SetModel The data is mined and extracted to obtain the data-driven analysis and processing model knowledge contained within these model sets. This knowledge is obtained by clustering and encapsulating the algorithms corresponding to information models and mechanism models of similar complex equipment. It includes the analysis and processing models corresponding to each node of the metastructure tree, and for supervised machine learning models, it also includes their labeled data. It can also be self-written data processing algorithms. Its characteristic is that the model works based on time-series-aware data and outputs analysis results, such as train bogie vibration processing models, axle temperature early warning models, and bearing life prediction models.

[0165] Preferably, in S412, the association rule knowledge contained in the data is mined and extracted. This type of knowledge comes from the mining of datasets in various fields of equipment and components. It is a rule obtained through association mining technology. For example, the fault association mining model for fault root cause analysis of key components such as train bearings, wheelsets, and suspension systems, and the rules obtained during the mining process.

[0166] Preferably, in S413, user intent recognition knowledge is mined and extracted from the data. This user intent recognition knowledge comes from mining datasets from various fields of equipment and components, as well as from unstructured knowledge extraction. This user intent recognition knowledge is used to fuzzily identify the user's intent and return results according to that intent, or to provide proactive push services to the user, including word segmentation models, intent recognition models, etc.

[0167] Preferably, in S414, external knowledge acquired through data collection is mined and extracted. This type of knowledge comes from Internet data in the field and is obtained through data collection and semi-structured knowledge extraction techniques.

[0168] Preferably, in S415, knowledge from relevant application software systems is mined and extracted. This type of knowledge comes from the knowledge accumulated by existing software systems at various stages of complex equipment, such as CAD requirements and objectives, CAPP process parameter libraries, etc.

[0169] Preferably, step S42 decomposes the complex equipment corresponding to the knowledge into structural dimensions. Dim 4 Layer Model Dimension Dim 3 M Time dimension Dim 1 T The integration process involves constructing a knowledge ontology, with specific implementation steps including S421-S423.

[0170] Preferably, step S421 defines the ontology, through Dim 5 Domain i= (Concept set, attribute set, relation set) complete the definition of the ontology and form a representation of knowledge.

[0171] Preferably, step S422 clarifies the object of the knowledge service and decomposes the structural dimensions. Dim 4 Layer To express.

[0172] Preferably, step S423 classifies various types of knowledge based on their service mechanisms and service types (such as annotation type, rule type, metadata type, and data-driven analysis and processing model type) and corresponding model dimensions. Dim 3 M Node time dimension Dim 1 T By mapping the nodes, a knowledge ontology is constructed that is jointly associated with the structural dimension, model dimension, and time dimension of complex equipment decomposition.

[0173] Preferably, step S43 involves ontology-level fusion and knowledge graph data-level fusion of knowledge from different domains. The specific implementation process includes steps S431-S432.

[0174] Preferably, step S431 achieves ontology fusion at the ontology level through concept and relation alignment and relation reasoning. Based on the similarity calculation of knowledge data attributes corresponding to concepts, the connections between concepts are reasoned, and numerical and textual attribute similarity measurement and overall concept similarity measurement algorithms are proposed, thereby achieving ontology-level fusion.

[0175] Preferably, the numerical and textual attribute similarity measurement algorithm in step S431 includes the following steps S4311-S4313.

[0176] Preferably, in step S4311, attribute normalization and entity clustering are performed, the units and names of attributes are unified, and a unit and constraint matching template is constructed. After the entities are classified according to the concepts in the ontology, similarity calculation is performed to achieve attribute alignment.

[0177] Preferably, in step S4312, similarity calculation is performed on numerical attributes (i.e., structured attributes), for example, for concepts. A Entities in a and concept B Entities in b To perform similarity calculation, first, for entities... a and b public attributes Similarity is calculated, and then weights are assigned to different structured attributes through methods such as manual scoring. The weight ratios of each attribute are obtained by weight normalization, and the overall structured attribute similarity is calculated.

[0178] Preferably, step S4313 calculates semantic similarity. For unstructured attributes (i.e. textual attributes) in an entity, the semantic similarity is calculated. First, the semantic feature vector of the textual attribute is constructed by word embedding through the description text of the corresponding attribute and the domain dictionary. Then, the similarity of the unstructured attribute is obtained by calculating the cosine similarity of the semantic feature vector.

[0179] Preferably, in step S432, after normalizing the weights of the structured attribute similarity and unstructured attribute similarity based on the overall concept similarity measurement algorithm in step S431 to obtain a comprehensive similarity, the following steps are performed for the concept... A Entities in a and concept B Entities in b Sort the candidates in descending order of similarity to generate the optimal candidate sequence, and select the entity with the highest similarity and a value greater than the threshold. b As an alignable entity.

[0180] Preferably, step S433 performs knowledge fusion at the knowledge graph data level. Building upon step S431, a unified knowledge representation model is established using Web Ontology Language (OWL). Knowledge graph embedding is achieved through word vectors, and the native graph database Neo4j is used for knowledge storage. For various knowledge sources and features, structured knowledge is fused from the database management system (DBMS) through Database to RDF (D2R) transformation. A BERT+Bilstm-CRF deep learning model is constructed to acquire unstructured text entities.

[0181] Preferably, for the extracted knowledge, entity alignment is performed on each clustered entity based on matching similarity and indicator graph embedding methods, thereby realizing knowledge fusion of the knowledge graph data layer; and the aggregation of AI model and algorithm knowledge is realized based on the AI ​​model set algorithm registry center.

[0182] Preferably, step S44 constructs a knowledge reasoning and intelligent service mechanism based on the knowledge types and application service needs of complex equipment. The specific implementation process includes steps S441-S443.

[0183] Preferably, step S441 involves sensing the request context of the user's requested knowledge and obtaining... Dim 4 Layer i , Dim 1 T , Dim 3 M The system then defines the scope of complex product domains, segments user requests into words, calculates the corresponding knowledge domains using word vectors, calls a domain intent recognition model to identify and transform user intents, and then queries the knowledge graph or calls the model algorithm to obtain query results and visualize them.

[0184] Preferably, step S442 performs sensor encoding acquisition and physical perception stream data analysis on the physical data-driven automatic request service, identifies the context and knowledge domain and obtains the knowledge domain encoding, and then generates the graph query basis, generates the AI ​​model and generates the function interface based on the knowledge domain encoding, and then queries the knowledge graph or calls the model algorithm to obtain the query results and displays them visually.

[0185] Preferably, in step S443, a knowledge proactive push service is constructed to perceive the user's current task and obtain the meta-model nodes of the equipment structure tree associated with the task. Dim 4 Layer ,Model Dim 3 M and corresponding life cycle stages Dim 1 T The system determines the scope of complex equipment domains, queries the knowledge graph for each knowledge domain and returns the knowledge dataset, generates domain labels, and then dynamically generates push window content.

[0186] In step S4, the knowledge mining, fusion, and service method for twin applications proposed in this invention outputs knowledge through data mining, knowledge extraction, ontology modeling, and ontology fusion, and then provides knowledge services for various types of twin applications through knowledge reasoning. This process indicates how to combine data, models, and metastructure trees through knowledge graphs in four-dimensional space to realize twin applications.

[0187] Preferably, step S5 includes: step S51: fusing the data of sibling nodes in the metastructure tree with the model set; step S52: fusing the data of parent and child nodes in the metastructure tree with the model; step S53: executing the digital twin application and outputting the results.

[0188] Preferably, in step S51, all data and models on the metastructure tree are first merged at the same level node. Data within the same node is merged in step S2, and models within the same node are merged in step S3. Intelligent computing and data mining are then performed in step S4 to form a mutual fusion of data and models within the same node.

[0189] Preferably, in step S52, the data and model associated with each node in the metastructure tree (such as the bogie node) are combined with the data and model of the parent node (such as the running gear node) and the child node (such as the wheelset node). The data combination includes the fusion of data between parent and child nodes through step S2 (such as the parent and child nodes formed by the running gear and the steering), while the model combination involves the fusion of models between parent and child nodes through step S3. Then, intelligent computing and data mining are performed through S4 to form a mutual fusion process of data and model between parent and child nodes.

[0190] Preferably, in step S53, the equipment data system, twin model, and knowledge are used as inputs. Existing historical data is processed through state judgment calculations, fault prediction calculations, correlation analysis, and data mining to form a knowledge graph representation of the knowledge model. Then, real-time data is used as input, and relevant knowledge models are retrieved from the knowledge graph based on data characteristics for calculation, enabling twin applications such as performance degradation analysis and lifespan prediction. Since twin applications are also composed of various algorithms, twins are defined as a set of algorithms. AlgorithmSet_Twin。

[0191] In step S5, the digital twin application proposed in this invention, based on the fusion of data dimension, structure dimension, and model dimension, defines the input, output, and algorithm set application paradigm of complex equipment digital twin application in four-dimensional space, providing a new approach to solving problems such as performance degradation analysis and fault diagnosis at each stage of the entire life cycle of complex equipment.

[0192] The above embodiments are only used to illustrate the present invention and are not intended to limit the technical solutions described herein. Although the present invention has been described in detail with reference to the above embodiments, the present invention is not limited to the specific embodiments described above. Therefore, any modifications or equivalent substitutions to the present invention, as well as all technical solutions and improvements that do not depart from the spirit and scope of the invention, are covered within the scope of the claims of the present invention.

Claims

1. A cyber-physical fusion method for the entire lifecycle of complex equipment, characterized in that: include: Step S1: Divide the complex equipment into four dimensions and multiple levels of structure; Step S2: Constructing a full lifecycle data system based on the fusion of data and time dimensions; Step S3: Constructing various types of models throughout the entire lifecycle based on the fusion of model dimension and time dimension; Step S4: Knowledge mining, fusion, and services for twin applications; Step S5: Digital twin application based on the multi-dimensional fusion of structural dimension, data dimension, and model dimension; Step S1 includes: Step S11: Four-dimensional definition of complex equipment; Step S12: Multi-level structure division; Step S13: Establishment of a coding system for complex equipment structure. In step S11: the four dimensions of complex equipment are defined as: time dimension Dim 1 T Data Dimension Dim 2 D Model dimension Dim 3 M Decomposition of structural dimensions Dim 4 Layer Specifically, it includes the following steps S111-S114; Step S111 defines the time dimension as Dim 1 T ,use Dim 1 T i This refers to different time phases throughout the lifecycle of complex equipment, including the design phase. Dim 1 T 1 Manufacturing stage Dim 1 T 2 and maintenance phase Dim 1 T 3 ; Step S112 defines the data dimension as Dim 2 D ;use Dim 2 D i This represents various types of data generated during the design, manufacturing, and operation and maintenance phases of complex equipment; the data from these three phases are processed through a dataset { Dim 2 D i SetData } to represent, i =1,2,3, where { Dim 2 D 1 SetData } represents the dataset for the design phase, { Dim 2 D 2 SetData } represents the manufacturing stage dataset, { Dim 2 D 3 SetData } represents the dataset for the operation and maintenance phase; Step S113 defines the model dimension as Dim 3 M ;use Dim 3 M i This represents the various models constructed during the design, manufacturing, and operation and maintenance phases of complex equipment; it also represents the models throughout the entire lifecycle of complex equipment as a model set { Dim 3 M i SetModel } to represent, i =1,2,3, where { Dim 3 M 1 SetModel } represents the design phase model set, { Dim 3 M 2 SetModel } represents the manufacturing stage model set, { Dim 3 M 3 SetModel } represents the set of models for the operation and maintenance phase; Step S114 defines the decomposition structure dimension as Dim 3 Layer ,use Dim 3 Layer i This indicates the structural composition of complex equipment; including the overall machine layer. Dim 3 Layer 1 System layer Dim 3 Layer 2 Subsystem layer Dim 3 Layer 3 Component layer Dim 3 Layer 4 Component layer Dim 3 Layer 5 ; Step S13 establishes a coding system for complex equipment structures, including complex equipment coding, complex equipment structure object coding, complex equipment lifecycle coding, complex equipment attribute parameter coding, and complex equipment lifecycle model coding. Step S13 includes steps S131-S135. Step S131 specifies the rules for the composition of complex equipment instance codes; the complex equipment instance code is defined as a code consisting of 5 groups of integers and strings, with the following structure: equipment code-major category-minor category-equipment type-equipment sequence number; Step S132 specifies the coding rules for complex equipment structure objects; the coding of complex equipment structure objects is defined as a code consisting of 7 groups of integers and strings, with the following structure: structure code-complete machine-system-subsystem-component-subcomponent-part; Step S133 specifies the life cycle coding of complex equipment; the life cycle stage coding of complex equipment is defined as a code consisting of 4 sets of integers and strings, with the following structure: complex equipment life cycle stage code - cycle stage - sub-stage - sub-process. Step S134 defines the encoding of complex equipment attribute parameters; the encoding of complex equipment attribute parameters is defined as an encoding consisting of 7 groups of integers and strings, with the following structure: code-category-major category-intermediate category-first-level classification-second-level classification-attribute parameter; Step S135 specifies the coding of the complex equipment life cycle model; the coding of the complex equipment life cycle model is defined as a code consisting of 4 sets of integers and strings, with the following structure: complex equipment life cycle model code - model classification - first-level sub-model - second-level sub-model. Step S2 includes: Step S21: Full lifecycle information physical data acquisition; Step S22: Source data processing and establishment of meta dataset; Step S23: Vertical fusion of data dimension and structural dimension; Step S24: Horizontal fusion of data dimension and time dimension. Step S22 includes steps S221-S223; In step S221, a meta-dataset is formed. The meta-dataset is a collection of data with a certain purpose or similarity formed by clustering different datasets or subsets. Step S222 analyzes the data format and processes the source data through a one-sided algorithm processing set to create standardized static and dynamic meta-datasets from the physical and information spaces. Step S223 fuses the metadata dataset by using a set of fusion algorithms to collect, process anomalies, and transform data to create a complete set of complex equipment metadata datasets. Step S23 associates the metadata datasets formed in each cycle of complex equipment design, manufacturing, and operation and maintenance with the hierarchical structural points of the equipment structure. The data association algorithm used is the association algorithm in the fusion algorithm set. With the metadata structure tree as the vertical axis, a vertical organization of the datasets with the metadata structure tree as the core is formed within each cycle stage. In the design stage, data vertical fusion is carried out based on the product prototype and the metadata structure tree organization. In the manufacturing stage, data vertical fusion is carried out based on product process planning and production organization. In the operation and maintenance stage, data vertical fusion is carried out based on product status monitoring, maintenance plan, and maintenance process organization. Step S24 is based on the coding system built on the metastructure tree after the digital prototype is formed. The mapping algorithm in the fusion algorithm set maps the design stage dataset to the next two cycle stages, forming a horizontal data system with time dimension as the core required by each stage, realizing horizontal data fusion. Horizontal data fusion is carried out from the design stage to the manufacturing stage, and the design data is transferred according to the product model data attached to each node of the metastructure tree. From the manufacturing stage to the operation and maintenance stage, the manufacturing data is transferred according to the manufacturing history attached to each node of the metastructure tree. From the design stage to the operation and maintenance stage, the design and manufacturing data are transferred according to the design and manufacturing history attached to each node of the metastructure tree. Step S3 includes: Step S31: Model composition and association at each stage; Step S32: Internal fusion of models at each stage to achieve dynamic updates; Step S33: Mutual fusion of multiple types of models to form twin models; Step S34: Mutual mapping and association of twin models at each stage. Step S4 includes: Step S41: Data mining and knowledge extraction; Step S42: Ontology modeling that integrates knowledge services and context; Step S43: Ontology fusion; Step S44: Knowledge reasoning and intelligent services. Step S5 includes: Step S51: Fusion of sibling node data in the metastructure tree with the model set; Step S52: Fusion of parent and child node data in the metastructure tree with the model; Step S53: Execution and result output of the digital twin application; In step S51, all data and models on the metastructure tree are first merged at the same level node. Data within the same node is merged in step S2, and models within the same node are merged in step S3. Intelligent computing and data mining are then performed in step S4 to form a mutual fusion of data and models within the same node. In step S52, the data and models associated with each node in the metastructure tree are combined with the data and models of the parent node and the child node. Data combination includes data fusion between parent and child nodes through step S2, while model combination involves model fusion between parent and child nodes through step S3. Then, intelligent computing and data mining are performed through S4 to form a mutual fusion process of data and models between parent and child nodes. In step S53, the equipment data system, twin model, and knowledge are taken as input. The existing historical data is used to form a knowledge model expressed by a knowledge graph through state judgment calculation, fault prediction calculation, correlation analysis, and data mining. Then, real-time data is taken as input, and relevant knowledge models are retrieved from the knowledge graph according to the data characteristics for calculation, and performance degradation analysis and life prediction twin applications are performed.

2. The cyber-physical fusion method for the entire lifecycle of complex equipment as described in claim 1, characterized in that: In step S12, a meta-structure tree is used as an index for the entire lifecycle process data and model to achieve consistent identification of data and model; the specific implementation process of step S12 includes the following steps S121-S124; Step S121 defines the meta-structure tree and the complex equipment decomposition structure dimension in S11; where each intermediate node on the meta-structure tree represents the complex equipment that can be further broken down into various subsystems and components; the construction of the minimum node of the meta-structure is the indivisible design component, the minimum assembly and manufacturing unit, and the minimum maintenance unit. Step S122 further divides the meta-structure tree into the design meta-structure tree, the manufacturing meta-structure tree, and the operation and maintenance meta-structure tree; Step S123 maps the metastructure tree to the bill of materials for the design, manufacturing, and operation and maintenance phases using a coding system; Step S124 divides the complex equipment into 6 levels from top to bottom, namely the complete machine level. Dim 4 Layer 1. System layer Dim 4 Layer 2 Subsystem layer Dim 4 Layer 3. Component layer Dim 4 Layer 4. Sub-component layer Dim 4 Layer 5. Component layer Dim 4 Layer 6.

3. The cyber-physical fusion method for the entire lifecycle of complex equipment as described in claim 2, characterized in that: Step S21 includes steps S211-S217; Step S211 defines the data sources for the entire lifecycle of complex equipment, including methods such as manual form filling, on-site data collection, and system generation; Step S212 defines the data types to be collected; Step S213 defines the data format to be collected; Step S214 defines the data format and corresponding acquisition method; Step S215 involves organizing the complex equipment's full lifecycle information physical data according to the data dimension composition format defined in step S112, and then combining the design phase dataset { Dim 2 D 1 SetData } Divide into design requirement dataset { Dim 2 D 1 SetDataRequirement Design a geometric dataset { Dim 2 D 1 SetDataGeom Design a physical dataset { Dim 2 D 1 SetDataPhysics }、Design process dataset{ Dim 2 D 1 SetDataProcess Design and analyze datasets { Dim 2 D 1 SetDataAnalyse } and design test dataset { Dim 2 D 1 SetDataTest }; Step S216 involves composing the complex equipment's full lifecycle information physical data according to the data dimension format defined in step S112, and then combining the manufacturing stage dataset { Dim 2 D 2 SetData } Divided into manufacturing planning dataset { Dim 2 D 2 SetDataPlan }, Manufacturing resource dataset { Dim 2 D 2 SetDataResource Manufacturing execution dataset { Dim 2 D 2 SetDataProc-exe Manufacturing quality dataset { Dim 2 D 1 SetDataQuality Manufacturing cost dataset { Dim 2 D 2 SetDataCost }; Step S217 will process the dataset for the operation and maintenance phase { Dim 2 D 3 SetData } is divided into operation and maintenance status datasets { Dim 2 D 3 SetDataStatus }、Operations and Maintenance Fault Dataset{ Dim 2 D 3 SetDataFault }、Operation and maintenance historical dataset{ Dim 2 D 3 SetDataRepairHistory }、Operation and maintenance cost dataset{ Dim 2 D 3 SetDataRepairCost }、Operations and Maintenance Plan Dataset{ Dim 2 D 3 SetRepairPlan }、 Operation and maintenance record dataset{ Dim 2 D 3 SetRepairRecord } 4. The cyber-physical fusion method for the entire lifecycle of complex equipment as described in claim 3, characterized in that: Step S31 includes steps S311-S317; Step S311 defines the composition of the design phase model set, which is the design information model set { in the design phase. Dim 3 M 1 SetInfoModel} Includes prototype models, requirement models, product models, and process models; design mechanism model set { Dim 3 M 1 SetPrinModel} Includes motion models, dynamic models, constitutive models, and analytical models; design domain model set { Dim 3 M 1 SetFeildModel} Includes whole-machine optimization model, component optimization model, and performance improvement model; Step S312 defines the composition of the manufacturing stage model set, in which the manufacturing information model set { Dim 3 M 2 SetInfoModel} It includes process models, cost models, quality models, and resource models; Manufacturing mechanism model set { Dim 3 M 2 SetPrinModel} It includes models of the production system's operating mechanism, equipment motion, functions, and performance; Manufacturing domain model set { Dim 3 M 2 SetFeildModel} It includes a production system control model, a quality analysis model, and an equipment operation and maintenance model; Step S313 defines the operation and maintenance phase model set { Dim 3 M 3 SetModel The composition of}, in the operation and maintenance phase, the operation and maintenance information model set { Dim 3 M 3 SetInfoModel It includes product models, maintenance plan models, knowledge models, and cost models; Operation and maintenance mechanism model set { Dim 3 M 3 SetPrinModel} Includes motion models, dynamic models, constitutive models, and analytical models; operation and maintenance domain model set { Dim 3 M 3 SetFeildModel} It includes fault prediction models, condition assessment models, life prediction models, and early warning models; Step S314 defines the design phase model set { Dim 3 M 1 SetModel The relationships between the three models contained within}; Step S315 defines the manufacturing stage model set { Dim 3 M 2 SetModel The relationships between the three models contained within}; Step S316 defines the operation and maintenance phase model set { Dim 3 M 2 SetModel The relationships between the three models contained within}; Step S317 defines the interaction relationships between the three phase model sets.

5. The cyber-physical fusion method for the entire lifecycle of complex equipment as described in claim 4, characterized in that: Step S33 includes steps S331-S333; Step S331: Based on the complex equipment's structural composition, working principle, operating environment, status, and twin application requirements, form an information model, mechanism model, and domain model encompassing the entire lifecycle. Through model classification, form model sets for each stage and type. Step S332, based on the application scenarios and needs of twin models, selects the appropriate models from the model set to construct twin models, and forms twin effects through data association between various types of models; Step S333 defines the role and process of the twin model in each stage. The twin role and process include design optimization and improvement in the design stage, production quality and production cost control in the manufacturing stage, and fault diagnosis, performance evaluation and analysis, and life prediction of each node in the operation and maintenance stage.

6. The cyber-physical fusion method for the entire lifecycle of complex equipment as described in claim 5, characterized in that: Step S34 includes steps S341-S343; In step S341, during the application of the twin model, the twin model is updated and optimized based on equipment history data analysis and knowledge graph services. Furthermore, for the complex equipment lifecycle business processes and activities, the interaction relationship between twin models is established through the mapping between twin models at different stages. Step S342 transfers the model generated in the design phase to the manufacturing phase through the meta-structure tree, and feeds back to the design phase through the role of the manufacturing process to update various models; Step S343 maps the manufacturing stage model to the operation and maintenance stage using a mapping algorithm from the model fusion algorithm set; Step S344 maps the design phase model to the operation and maintenance phase to understand the performance, remaining lifespan, etc. of each node on the equipment metastructure tree. The results of the operation and maintenance phase twin model will also be fed back to the design phase, forming the demand points for design improvement and driving the closed-loop feedback of the entire life cycle of the complex equipment.

7. The cyber-physical fusion method for the entire lifecycle of complex equipment as described in claim 6, characterized in that: Step S41 includes steps S411-S415; In S411, for the model dimension Dim 3 M We mine and extract various model sets of complex equipment in the design, manufacturing and operation and maintenance stages to obtain the data-driven analysis and processing model class knowledge contained in these model sets. In S412, association rule knowledge contained in the data is mined and extracted. This type of knowledge comes from the mining of datasets in various fields of equipment and parts. In S413, user intent recognition knowledge is mined and extracted from the data. This user intent recognition knowledge comes from the mining of datasets in various fields of equipment and parts, as well as the extraction of unstructured knowledge. User intent recognition knowledge is used to fuzzily identify user intent and return results according to user intent, or to provide proactive push services to users, including word segmentation processing models and intent recognition models; In S414, external knowledge acquired through data collection is mined and extracted. This type of knowledge comes from Internet data in the field and is obtained through data collection and semi-structured knowledge extraction techniques. In S415, knowledge from relevant application software systems is mined and extracted. This type of knowledge comes from the knowledge accumulated by existing software systems at various stages of complex equipment.

8. The cyber-physical fusion method for the entire lifecycle of complex equipment as described in claim 7, characterized in that: Step S43 includes steps S431-S432 Step S431 achieves ontology fusion at the ontology level through concept, relation alignment, and relation reasoning. Based on the similarity calculation of knowledge data attributes corresponding to concepts, it infers the connection between concepts and proposes numerical and textual attribute similarity measurement and overall concept similarity measurement algorithms, thereby achieving ontology-level fusion. In step S432, the structured attribute similarity and unstructured attribute similarity are weighted and normalized according to the concept overall similarity measurement algorithm in step S431 to obtain the comprehensive similarity. Then, the optimal candidate sequence is generated, and the entity with the highest similarity and greater than the threshold is selected as the entity that can be aligned. Step S433 performs knowledge fusion at the knowledge graph data level; Based on step S431, a unified knowledge representation model is established using a network ontology language. Knowledge graphs are embedded through word vectors, and knowledge is stored using a native graph database. For various knowledge sources and features, structured knowledge is integrated from the database management system through D2R transformation. A deep learning model of BERT+Bilstm-CRF is constructed to obtain unstructured text entities.

Images