A Method for Constructing Multi-Source Heterogeneous Data Fusion Patterns for the Entire Lifecycle of Complex Equipment

Abstract

This invention relates to a method for constructing a multi-source heterogeneous data fusion model for the entire lifecycle of complex equipment. The method includes: establishing a method for identifying redundant attributes of multi-source heterogeneous data for complex equipment; establishing algorithms for reusable knowledge data association mapping based on product family metastructure trees, generating design structure trees and data association mapping for the design stage, generating manufacturing structure trees and data association mapping for the manufacturing stage, generating operation and maintenance structure trees and data association mapping for the operation and maintenance stage; and constructing a data model for the entire lifecycle of complex equipment that integrates multi-source heterogeneous data based on functional dependencies and foreign key constraints of relational models. This invention solves the problems of data silos caused by the differences in physical information data structures, feature redundancy, and lack of data relationships resulting from the independent application of various business systems in the design, manufacturing, and operation and maintenance stages of complex equipment by establishing data association mapping and inter-stage data mapping methods for the design, manufacturing, and operation and maintenance stages of complex equipment.

Description

TECHNICAL FIELD

[0001] The application relates to a physical information data fusion mode construction method, in particular to a complex equipment whole life cycle multi-source heterogeneous data fusion mode construction method. BACKGROUND

[0002] Complex equipment such as high-speed trains, shield machines and nuclear power equipment has a design, manufacturing and operation process that spans multiple professions and enterprises and involves numerous physical information systems. Due to the different problems and providers of each physical information system, the multi-source heterogeneity of complex equipment whole life cycle data makes the multi-dimensional heterogeneous information in different disciplines and different stages of the product life cycle unable to be uniformly expressed at the system level. Current research focuses mostly on general multi-source heterogeneous data conversion and fusion algorithms, multi-source multi-dimensional heterogeneous multi-modal complex data classification and modeling methods, and no research has been conducted on data correlation mapping in each stage of complex equipment design, manufacturing and operation and stage-to-stage data correlation mapping methods and integrated data mode construction methods.

[0003] The collection, correlation mapping and closed-loop feedback of complex equipment whole life cycle data are the cornerstone of building design optimization iteration and digital twinning. Through data mining and intelligent computing, the parameters of each stage can be effectively improved, providing important value for continuous optimization and iteration of products. However, the multi-source heterogeneity of a large amount of physical information data accumulated in the current stages makes the data lack completeness and consistency, greatly affecting the credibility and usability of the data, making it difficult to effectively realize the value of the data. SUMMARY

[0004] In order to achieve the above application purposes, the application provides the following technical solutions:

[0005] A complex equipment whole life cycle multi-source heterogeneous data fusion mode construction method comprises the following steps:

[0006] Step S1: identification and reduction of multi-source heterogeneous data redundant attributes;

[0007] Step S2: establishment of a correlation mapping model between data entities;

[0008] Step S3: correlation mapping of product family meta-structure trees and knowledge data;

[0009] Step S4: design stage data correlation mapping;

[0010] Step S5: manufacturing stage data correlation mapping;

[0011] Step S6: operation and maintenance stage data correlation mapping;

[0012] Step S7: multi-source heterogeneous data conversion and fusion storage method;

[0013] Wherein, step S1 comprises step S11-step S12;

[0014] Step S11: multi-source heterogeneous data main attribute marking;

[0015] For two data sources D x1 and D x2 of the same data object X, different redundant attribute identification and reduction methods are proposed for nominal attributes and numerical attributes respectively; first, read the database dictionary to obtain the respective main attributes PK(D x1 ) and PK(D x2 ), judge whether PK(D x1 ) and PK(D x2 ) are globally unique identifiers or self-increment types generated by the algorithm, and handle the following 3 cases:

[0016] (1) If one of them is a globally unique identifier or a self-increment type, mark the attribute for deletion, and mark the other attribute as the main attribute;

[0017] (2) If both are globally unique identifiers or self-increment types, remind the administrator to select the main attribute from the remaining attributes of D x1 and D x2 , and mark the two main attributes for deletion;

[0018] (3) If neither is a globally unique identifier nor a self-increment type, judge whether the names and data types of the two are the same, if the same, mark the one with more data samples as the main attribute, if different, remind the administrator to select one from PK(D x1 ) and PK(D x2 ) to mark as the main attribute;

[0019] Step S12: divide the attributes into nominal attributes and numerical attributes according to the attribute value range, and reduce the attributes belonging to the same data type; step S12 comprises step S121-step S122;

[0020] Step S121: nominal attribute reduction;

[0021] Attributes A and B are two attributes from different systems D x1 and D x2 , attribute A has c values A(a1, a2, …, a c ), and attribute B has r values B(b1, b2, …, b r ), then whether attributes A and B describe the same feature is calculated as formula (1):

[0022]

[0023] Where: x2 is the chi-square test value, o ij is the observed frequency, e ij is the expected frequency, e ij The calculation formula (2) is as follows:

[0024]

[0025] count (A=a i ) represents the number of data samples taking a i value, count (B=b j ) is the number of samples taking b j value, and n is the total number of data samples; x 2 calculated is compared through the chi-square table, and if relevant, the data type definition length of the properties A and B is compared, the property with larger data type definition length is retained, and the other property is marked as redundant;

[0026] Step S122: numerical property reduction;

[0027] The properties A and B are two properties from different systems, and the correlation degree of the properties A and B is measured by calculating the covariance as shown in formula (3):

[0028]

[0029] In the formula, a i ∈{a1,a2,…,a n ) is the value of n samples of the property A, b i ∈{b1,b2,…,b n ) is the value of n samples of the property B, and are the mean values of the properties A and B respectively, and n is the sample number;

[0030] The correlation coefficient of the properties A and B is further calculated from the covariance as shown in formula (4):

[0031]

[0032] In the formula, σ A and σ B represent the standard deviations of the data corresponding to the properties A and B respectively;

[0033] The calculation result r A,B is compared with the set threshold value, and if it exceeds the threshold value, then:

[0034] The data type definition length of the properties A and B is compared, the property with larger data type definition length is retained, and the other property is marked as redundant;

[0035] Step S2: Establish a mapping model for the association between data entities;

[0036] Let D x and D y The data entities representing two different types of objects are mapped using a relational data model to establish referential integrity relationships between these two types of data entities, as shown in equation (5):

[0037]

[0038] In the formula: f(D) x D y ) represents D x and D y The correlation mapping function between them, D x ×D y D represents x and D y Perform a Cartesian product operation on two datasets. Indicated in D x The A and D attributes of the data schema y In the conditional selection operation where the A' attribute values ​​of the data schema are equal, in the above mapping relationship, D x This is called a reference data relation entity, D x A is D x The reference attribute, D y The entity referred to as the referenced entity, D y A' is D y The set of primary attributes;

[0039] Step S7 includes steps S71-S74;

[0040] Step S71: For data object X, the data schema of business system J is as follows: For a given set of n attributes The data pattern Its corresponding full lifecycle data warehouse model S x (A x1 A x2 ,…A xm ), which consists of m attributes A x1 A x2 ,…A xm The data pattern constitutes the judgment of A x1 A x2 ,…A xm If there are any reference attributes established by the associated mapping, then process them as follows:

[0041] If there is a reference attribute: check whether the data object data referred to has been imported; if not imported: remind to import the reference entity data object first and end; if imported: judge whether the reference attribute data set is in the reference entity attribute, and if not all, remind the user of the reference data anomaly;

[0042] Step S72: according to the similarity calculation mark in S1, remove the marked attributes, and the remaining k attributes Matching by name or redundancy mark, matching with S x Mapping of k attributes

[0043] Step S3 includes steps S31-S32;

[0044] Step S31: construct the self-association mapping between product family meta-structure tree node data;

[0045] Step S32: establish the mapping between product family meta-structure tree nodes and family model, rule knowledge data.

[0046] Step S4 includes steps S41-S42;

[0047] Step S41: construct the mapping between product design structure tree and meta-structure tree nodes;

[0048] Step S41 includes steps S411-S412;

[0049] Step S411: establish the self-association mapping of design structure tree node data;

[0050] Step S412: establish the association mapping between design structure tree nodes and meta-structure tree nodes;

[0051] Step S42: establish the mapping between design structure tree nodes and design stage design model, design parameter, simulation analysis, strength analysis design stage data.

[0052] Step S5 includes steps S51-S52;

[0053] Step S51: construct the mapping between product manufacturing structure tree and design structure tree nodes;

[0054] Wherein, step S51 includes steps S511-S512;

[0055] Step S511: establish the self-association mapping of manufacturing structure tree node data;

[0056] Step S512: establish the association mapping between manufacturing structure tree nodes and design structure tree nodes;

[0057] Step S52: Establish the mapping between manufacturing structure tree nodes and manufacturing stage process routes, operations, production tasks, task scheduling, production execution reporting, and quality inspection data.

[0058] Step S6 includes steps S61-S62;

[0059] Step S61: Construct a node mapping between the product operation and maintenance structure tree and the design structure tree or manufacturing structure tree;

[0060] Step S611: Establish a self-associative mapping of node data in the operation and maintenance structure tree;

[0061] Step S612: Establish the association mapping between operation and maintenance structure tree nodes and manufacturing structure tree nodes or design structure tree nodes;

[0062] Step S62: Establish a mapping between product operation and maintenance structure tree nodes and installation files, operation and maintenance records, perception data, fault data, and repair and replacement data during the operation and maintenance phase.

[0063] Step S73: For each attribute pair examine and A xq Check if the data type and unit are consistent; if not, convert them.

[0064] Step S73 includes steps S731-S733;

[0065] Step S731: If the data types are inconsistent but both are numeric, then call the unit conversion function in equation (6):

[0066]

[0067] In the formula: for Data type, A xq .DataType is A xq The data type, this function is based on Data types and A xq The data type is determined by calling the corresponding function in the type conversion library to perform the conversion.

[0068] Step S732: If the dimensions are inconsistent, call the dimension conversion function of equation (7):

[0069]

[0070] In the formula: for Data units, A xq .Dimension is A xq The data units of this function are based on Data dimensions and Axq The data dimension is converted by calling a function of a type conversion library.

[0071] Step S733: writing data into the data warehouse x .

[0072] Step S74: repeating the processes of S71-S73 on the other source of the data object X, and only updating the hollow value attribute for the data with the same main attribute value, without repeated writing. x .

[0073] Compared with the prior art, the present application has the following beneficial effects:

[0074] (1) The inventor found in practice that in the prior art data integration method, the modeling personnel needs to manually analyze and extract the characteristics of the multi-source heterogeneous data sources, and construct a set of characteristic attribute methods for each type of data. The process is time-consuming and relies too much on the experience of the modelers. The present application designs a main attribute discrimination and elimination method for each type of data from multiple sources, calculates the similarity of each attribute through data mining, marks the redundant attributes, and automatically constructs the characteristic attribute set of each type of data entity.

[0075] (2) The inventor found in practice that some business systems have formed reusable knowledge data such as professional family library, design rules, and maintenance methods in application. Such data is very important for establishing the correlation between products and tracing the source. The present application establishes the association mapping between knowledge data and meta-structure tree nodes through the product family meta-structure tree, and establishes the join dependency relationship between knowledge data through the establishment of data reference integrity constraints.

[0076] (3) The inventor found in practice that the data generated in the complex equipment design process corresponds to a node of the product design structure tree, and each node of the design structure tree can be obtained by instantiating and mapping the meta-structure tree node. For this, the present application proposes a method for associating and mapping the data of the meta-structure tree and the design structure tree, as well as the data of the design stage parts model, two-dimensional atlas, simulation analysis, strength analysis, and the like with the nodes of the design structure tree and the family model. The join dependency relationship of the design stage data is established through data reference integrity mapping.

[0077] (4) The inventor found in practice that the data generated in the complex equipment manufacturing process corresponds to a node of the product manufacturing structure tree, and the manufacturing structure tree can be generated by mapping the design structure tree. For this, the present application proposes a mapping method between the nodes of the manufacturing structure tree and the nodes of the design structure tree, and a method for associating and mapping the data of the manufacturing stage production process, process, production task, task scheduling, production report, and quality inspection with the manufacturing structure tree. The join dependency relationship of the data is established through data reference integrity mapping, thereby associating the data of the manufacturing and design stages.

[0078] (5) The inventor finds in practice that the data generated in the complex equipment operation and maintenance process correspond to a node of the product operation and maintenance structure tree, and the node of the operation and maintenance structure tree can be obtained from the manufacturing structure tree or the design structure tree node mapping. In this regard, the application proposes the mapping relationship between the operation and maintenance structure tree node and the design and manufacturing structure tree node, and establishes the associated mapping between the operation and maintenance structure tree node and the installation file, operation and maintenance record, sensing data, fault alarm, maintenance and replacement parts, etc. The data connection dependency relationship is established through data reference integrity mapping, so as to associate the data of each stage of operation, manufacturing and design, and form a full life cycle data system.

[0079] (6) The inventor finds in practice that there are data type conversion and data specification problems in the integration of each heterogeneous data into a unified data mode in multi-source heterogeneous data fusion. In this regard, the application abstracts the data conversion type, establishes a multi-source heterogeneous data conversion and data specification fusion algorithm, and solves the multi-source heterogeneous data extraction and conversion problem. BRIEF DESCRIPTION OF DRAWINGS

[0080] Figure 1 It is a schematic diagram of the overall solution;

[0081] Figure 2 It is a product family meta-structure tree and knowledge class data mode associated mapping schematic diagram;

[0082] Figure 3 It is a product design structure tree and design stage data mode associated mapping schematic diagram;

[0083] Figure 4 It is a product manufacturing structure tree and manufacturing stage data mode associated mapping schematic diagram;

[0084] Figure 5 It is a product operation and maintenance structure tree and operation and maintenance stage data mode associated mapping schematic diagram. DETAILED DESCRIPTION

[0085] In order to make the purpose, technical scheme and advantages of the embodiments of the application clearer, the technical scheme in the embodiments of the application will be described clearly and completely below with reference to the drawings. Obviously, the described embodiments are part of the embodiments of the application, rather than all the embodiments.

[0086] Therefore, the following detailed description of the embodiments of the application is not intended to limit the scope of the claimed application, but only represents some embodiments of the application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the application without creative labor are within the scope of protection of the application.

[0087] It should be noted that the embodiments in the application and the features and technical solutions in the embodiments can be combined with each other without conflict.

[0088] A complex equipment whole life cycle multi-source heterogeneous data fusion mode construction method, comprising the following steps:

[0089] Step S1: identification and reduction of redundant attributes of multi-source heterogeneous data.

[0090] Preferably, the characteristics of the data are described by attributes, and the attribute names of different source systems for the same characteristics of the same type of data are different, and direct attribute merging will cause attribute explosion and data inconsistency problems. In view of this, the application proposes a redundant attribute elimination method based on multi-source data. For two data sources D x1 and D x2 of the same data object X, different redundant attribute identification and reduction methods are proposed for nominal attributes and numerical attributes.

[0091] Preferably, step S1 includes steps S11-S12.

[0092] Preferably, step S11: multi-source heterogeneous data primary attribute marking.

[0093] Read the database dictionary to obtain the respective primary attributes PK(D x1 ) and PK(D x2 ), and judge whether PK(D x1 ) and PK(D x2 ) are globally unique identifiers or self-increment types generated by the algorithm, and handle the following cases:

[0094] (1) If one of them is a globally unique identifier or a self-increment type, mark the attribute for deletion, and mark the other attribute as the primary attribute.

[0095] (2) If both are globally unique identifiers or self-increment types, remind the administrator to select the primary attribute from the remaining attributes of D x1 and D x2 , and mark the two primary attributes for deletion.

[0096] (3) If neither is a globally unique identifier nor a self-increment type, judge whether the names and data types of the two are the same. If they are the same, mark the one with more data samples as the primary attribute. If they are different, remind the administrator to select one from PK(D x1 ) and PK(D x2 ) to mark as the primary attribute.

[0097] Preferably, step S12: redundant attribute identification method.

[0098] According to the attribute value range, the attributes are divided into nominal attributes and numerical attributes, and the attributes belonging to the same data type are reduced, respectively as follows:

[0099] Preferably, step S121: nominal attribute reduction.

[0100] Attribute A and attribute B are two attributes from different systems D x1 and D x2 , attribute A has c values, A (a1, a2, …, a c ), and attribute B has r values B (b1, b2, …, b r ), whether attribute A and attribute B describe the same feature is calculated as follows:

[0101]

[0102] Where: x 2 is the chi-square test value, o ij is the observed frequency, e ij is the expected frequency, and the calculation formula of e ij is as follows:

[0103]

[0104] count (A = a i ) represents the number of data samples with value a i , count (B = b j ) is the number of samples with value b j , and n is the total number of data samples. For the calculated x 2 , compare it through the chi-square table. If it is related, compare the data type definition length of attribute A and B, keep the attribute with the longer data type definition length, and mark the other attribute as redundant.

[0105] According to the above method, test the nominal attributes from different systems two by two, and complete the marking of all redundant nominal type attributes.

[0106] Preferably, step S122: numerical attribute reduction.

[0107] Attribute A and attribute B are two attributes from different systems, and the correlation between attribute A and B is measured by calculating the covariance, that is:

[0108]

[0109] In the formula, a i ∈{a1, a2, …, a n ) is the value of the n samples of attribute A, b i ∈{b1, b2, …, b n ) is the value of the n samples of attribute B, and are the mean values of attributes A and B respectively, and n is the sample number.

[0110] The correlation coefficient of attributes A and B is further calculated by covariance:

[0111]

[0112] In the formula, σ A and σ B respectively represent the standard deviation of the data corresponding to attributes A and B.

[0113] The calculation result r A,B is compared with the set threshold value, and if it exceeds the threshold value, then:

[0114] The data type definition length of attributes A and B is compared, the attribute with the larger data type definition length is retained, and the other attribute is marked as redundant.

[0115] According to the above method, the numerical attributes from different systems are tested pairwise, and the redundant attribute marking of all numerical types is completed.

[0116] In view of the fact that in the prior art data integration method, the modeling personnel needs to manually analyze and extract the characteristics of the multi-source heterogeneous data sources, and constructs a set of characteristic attribute methods for each type of data, the process is time-consuming and relies too much on the experience of the modeler, the present application designs a main attribute discrimination and elimination method for each type of data from multiple sources, calculates the similarity of each attribute through data mining, marks the redundant attributes, and automatically constructs the characteristic attribute set of each type of data entity. Through the method of S1, multiple data sources are marked and identified one by one, and the construction of the same data entity data mode is completed.

[0117] Preferably, step S2: establishing a correlation mapping model between data entities.

[0118] Let D x and D y represent two types of data entities of different objects, and a relational data model is used to establish the referential integrity correlation mapping between the two types of data entities:

[0119]

[0120] In the formula, f(D x ,D y ) represents the correlation mapping function between D x and D y , D x x D y represents the Cartesian product operation of D x and D y , represents the conditional selection operation under the condition that the A attribute of D x data mode and the A' attribute value of D y data mode are equal, and in the above mapping relationship, Dx D x .A is called the reference attribute of D x D y D y .A' is called the primary attribute set of D y .

[0121] Formula (5) can be achieved by the foreign key constraint of the relationship model, and converted into the relational database definition language as follows: CONSTRAINT constraint name FOREIGN KEY D x (A) REFERNECES D y (A'), that is, adding a reference attribute on D x to associate D y , so as to realize the reference integrity verification through the database.

[0122] Preferably, step S3: product family meta-structure tree and knowledge class data association mapping.

[0123] Preferably, step S31: self-association mapping between product family meta-structure tree node data is constructed.

[0124] The assembly parent-child integrity constraint association mapping between the product family meta-structure tree nodes is as follows by using formula (5):

[0125] f (meta-structure tree node, meta-structure tree node) = σ 元结构树节点.父节点类编码＝元结构树节点.节点类编码 (meta-structure tree node x meta-structure tree node).

[0126] Step S32: mapping between product family meta-structure tree nodes and family model, rule and other knowledge data is established.

[0127] The integrity constraint association mapping of each family model belonging to a meta-structure tree node is as follows by using formula (5):

[0128] f (family model, meta-structure tree node) = σ 族模型.节点类编码＝元结构树节点.节点类编码 (family model x meta-structure tree node).

[0129] The integrity constraint association mapping of each parameter belonging to a family model is as follows by using formula (5):

[0130] f (parameter, family model) = σ 参数.族模型编码＝族模型.族模型编码 (parameter x family model).

[0131] The integrity constraint association mapping of each rule applicable to a meta-structure tree node is as follows by using formula (5):

[0132] f (rule, meta-structure tree node) = σ 规则.节点类编码＝元结构树节点.节点类编码 (rule x meta-structure tree node).

[0133] The integrity constraint association for each rule that applies to one lifecycle phase is mapped using equation (5) as follows:

[0134] f (rule, lifecycle phase) = σ 规则.阶段编码＝生命周期阶段.阶段编码 (rule x lifecycle phase).

[0135] The integrity constraint association for each criterion that applies to one or more meta-structure tree nodes is mapped using equation (5) as follows:

[0136] f (criterion, meta-structure tree node) = σ 标准.节点类编码＝元结构树节点.节点类编码 (criterion x meta-structure tree node).

[0137] The integrity constraint association for each criterion that applies to one or more lifecycle phases is mapped using equation (5) as follows:

[0138] f (criterion, lifecycle phase) = σ 标准.阶段编码＝生命周期阶段.阶段编码 (criterion x lifecycle phase).

[0139] The integrity constraint association for each criterion item that describes a requirement for some aspect of a parameter is mapped using equation (5) as follows:

[0140] f (criterion item, parameter) = σ 标准项.参数编码＝参数.参数编码 (criterion item x parameter).

[0141] The integrity constraint association for each criterion item that comes from a criterion is mapped using equation (5) as follows:

[0142] f (criterion item, criterion) = σ 标准项.标准编码＝标准.标准编码 (criterion item x criterion).

[0143] The integrity constraint association for each data-driven analytic processing model that serves one meta-structure tree node is mapped using equation (5) as follows:

[0144] f (data-driven analytic processing model, meta-structure tree node) = σ

[0145] σ 数据驱动的分析处理模型.节点类编码＝元结构树节点.节点类编码 (data-driven analytic processing model x meta-structure tree node).

[0146] The integrity constraint association for each data-driven analytic processing model that applies to one lifecycle phase is mapped using equation (5) as follows:

[0147] f (data-driven analytic processing model, lifecycle phase) = σ

[0148] σ 数据驱动的分析处理模型.阶段编码＝生命周期阶段.阶段编码 (data-driven analytic processing model x lifecycle phase).

[0149] The integrity constraint association mapping of each field dictionary belonging to a meta-structure tree node is as follows by using formula (5):

[0150] f(field dictionary, meta-structure tree node) = σ 领域字典.节点类编码＝元结构树节点.节点类编码 (field dictionary x meta-structure tree node).

[0151] The integrity constraint association mapping of each item of external knowledge belonging to a meta-structure tree node is as follows by using formula (5):

[0152] f(external knowledge, meta-structure tree node) = σ 外部知识.节点类编码＝元结构树节点.节点类编码 (external knowledge x meta-structure tree node).

[0153] The integrity constraint association mapping of each item of external knowledge belonging to a life cycle stage is as follows by using formula (5):

[0154] f(external knowledge, life cycle stage) = σ 外部知识.阶段编码＝生命周期阶段.阶段编码 (external knowledge x life cycle stage).

[0155] In view of the fact that part of business systems have formed reusable knowledge data such as professional family library, design rules, maintenance methods and the like in application, the data is very important for establishing the correlation relationship between products, the application establishes the association mapping between the knowledge data and the meta-structure tree node through the product family meta-structure tree, and establishes the coupling dependency relationship between the knowledge data through the establishment of data reference integrity constraint.

[0156] Step S4: design stage data association mapping.

[0157] Step S41: mapping of product design structure tree and meta-structure tree node is constructed.

[0158] Step S411: self-association mapping of design structure tree node data is established.

[0159] The integrity constraint association mapping of the assembly parent-child relationship of the product design structure tree node is as follows by using formula (5):

[0160] f (design structure tree node, design structure tree node) = σ 设计结构树节点.父节点编码＝设计结构树节点.节点编码 (design structure tree node x design structure tree node).

[0161] Step S412: association mapping between design structure tree node and meta-structure tree node is established.

[0162] The integrity constraint association mapping of the inheritance of the product design structure tree node from the meta-structure tree node is as follows by using formula (5):

[0163] f (design structure tree node, meta-structure tree node) =

[0164] σ设计结构树节点.节点类编码＝元结构树节点.节点类编码 (Design Structure Tree Node x Meta Structure Tree Node).

[0165] Step S42: Establish the mapping of the design structure tree node and the design stage design model, design parameter, simulation analysis, strength analysis and other design stage data as follows:

[0166] The integrity constraint association relationship between each design model belonging to a product design structure tree node is mapped as follows using formula (5):

[0167] f (Design Model, Design Structure Tree Node) = σ 设计模型.节点编码＝设计结构树节点.节点编码 (Design Model x Design Structure Tree Node).

[0168] The instantiation reference integrity constraint between the design model and the family model is mapped as follows using formula (5):

[0169] f (Design Model, Family Model) = σ 设计模型.族模型编码＝族模型.族模型编码 (Design Model x Family Model).

[0170] The integrity constraint association between each design parameter belonging to a design model is mapped as follows using formula (5):

[0171] f (Design Parameter, Design Model) = σ 设计参数.模型编码＝设计模型.模型编码 (Design Parameter x Design Model).

[0172] The instantiation integrity constraint association of each design parameter corresponding to the parameter is mapped as follows using formula (5):

[0173] f (Design Parameter, Parameter) = σ 设计参数.参数类编码＝参数.参数编码 (Design Parameter x Parameter).

[0174] The integrity constraint association of each requirement data corresponding to one or more design structure tree nodes is mapped as follows using formula (5):

[0175] f (Requirement Data, Design Structure Tree Node) = σ 需求数据.节点编码＝设计结构树节点.节点编码 (Requirement Data x Design Structure Tree Node).

[0176] The integrity constraint association of each simulation analysis data corresponding to a design model is mapped as follows using formula (5):

[0177] f (Simulation Analysis, Design Model) = σ 仿真分析.模型编码＝设计模型.模型编码 (Simulation Analysis x Design Model).

[0178] The integrity constraint association of each strength analysis data corresponding to a design model is mapped as follows using formula (5):

[0179] f (Strength Analysis, Design Model) = σ 强度分析.模型编码＝设计模型.模型编码(Strength analysis x design model).

[0180] The integrity constraint correlation mapping of each two-dimensional drawing corresponding to a design model is as follows by using formula (5):

[0181] f (two-dimensional drawing, design model) = σ 二维图册.模型编码＝设计模型.模型编码 (two-dimensional drawing x design model).

[0182] The constraint relationship of each two-dimensional drawing belonging to a design structure tree node is as follows by using formula (5):

[0183] f (two-dimensional drawing, design structure tree node) = σ 二维图册.节点编码＝设计结构树节点.节点编码 (two-dimensional drawing x design structure tree node).

[0184] The integrity constraint correlation mapping of each two-dimensional drawing completed by a designer is as follows by using formula (5):

[0185] f (two-dimensional drawing, employee) = σ 二维图册.工号＝设计员.工号 (two-dimensional drawing x employee).

[0186] The integrity constraint correlation mapping of each design review record corresponding to a design model is as follows by using formula (5):

[0187] f (design review, design model) = σ 设计评审.模型编码＝设计模型.模型编码 (design review x design model).

[0188] The integrity constraint correlation mapping of each design model completed by a designer is as follows by using formula (5):

[0189] f (design model, employee) = σ 设计模型.工号＝员工.工号 (design model x employee).

[0190] In view of the fact that the data generated in the complex equipment design process correspond to a node of the product design structure tree, and each node of the design structure tree can be obtained by instantiation mapping from the meta-structure tree node, the application proposes a method for data correlation mapping between the meta-structure tree and the design structure tree, and for correlation mapping between the design stage part model, two-dimensional drawing, simulation analysis, strength analysis and other data and the design structure tree node and family model, and the coupling dependency relationship of the design stage data is established through data reference integrity mapping.

[0191] Preferably, step S5: manufacturing stage data correlation mapping.

[0192] Preferably, step S51: constructing a product manufacturing structure tree and a design structure tree node mapping.

[0193] Preferably, step S511: establishing a self-correlation mapping of the manufacturing structure tree node data.

[0194] The integrity constraint association mapping of the product manufacturing structure tree node assembly parent-child relationship is mapped by formula (5) as follows:

[0195] f(manufacturing structure tree node, manufacturing structure tree node) = σ

[0196] σ 制造结构树节点.父节点编码＝制造结构树节点.节点编码 (manufacturing structure tree node x manufacturing structure tree node).

[0197] Preferably, step S512: establish the association mapping between the manufacturing structure tree node and the design structure tree node.

[0198] The integrity constraint association mapping of the product manufacturing structure tree node inherited from the design structure tree node is mapped by formula (5) as follows:

[0199] f(manufacturing structure tree node, design structure tree node) = σ 制造结构树节点.设计节点编码＝设计结构树节点.节点编码 (manufacturing structure tree node x design structure tree node).

[0200] Step S52: establish the mapping of the manufacturing structure tree node and the manufacturing stage process route, process, production task, task scheduling, production execution report, quality detection and other data.

[0201] The integrity constraint association mapping of each process route belonging to a design structure tree node is mapped by formula (5) as follows:

[0202] f(process route, design structure tree node) = σ 工艺路线.节点编码＝设计结构树节点.节点编码 (process route x design structure tree node).

[0203] The integrity constraint association mapping of each process belonging to a process route is mapped by formula (5) as follows:

[0204] f(process, process route) = σ 工序.工艺编码＝工艺路线.工艺编码 (process x process route).

[0205] The integrity constraint association mapping of each production task corresponding to a manufacturing structure tree node is mapped by formula (5) as follows:

[0206] f(production task, manufacturing structure tree node) = σ 生产任务.智造码＝制造结构树节点.制造码 (production task x manufacturing structure tree node).

[0207] The integrity constraint association mapping of each production task corresponding to a process route is mapped by formula (5) as follows:

[0208] f(production task, process route) = σ 生产任务.工艺编码＝工艺路线.工艺编码 (production task x process route).

[0209] The integrity constraint association of each production job belonging to a production task is mapped as follows by using formula (5):

[0210] f(production job, production task) = σ 生产作业.任务编码＝生产任务.任务编码 (production job x production task).

[0211] The integrity constraint association of each quality inspection job corresponding to a production job is mapped as follows by using formula (5):

[0212] f(quality inspection job, production job) = σ 质检作业.作业编码＝生产作业.作业编码 (quality inspection job x production job).

[0213] The integrity constraint association of each quality inspection record corresponding to a quality inspection job is mapped as follows by using formula (5):

[0214] f(quality inspection record, quality inspection job) = σ 质检记录.作业编码＝质检作业.作业编码 (quality inspection record x quality inspection job).

[0215] The integrity constraint association of each production job being completed by a production service unit is mapped as follows by using formula (5):

[0216] f(production job, production service unit) = σ 生产作业.服务单元编码＝生产服务单元.服务单元编码 (production job x production service unit).

[0217] The integrity constraint association of each production job using a tooling must belong to the tooling information is mapped as follows by using formula (5):

[0218] f(production job, tooling information) = σ 生产作业.工装编码＝工装信息.工装编码 (production job x tooling information).

[0219] The integrity constraint association of each production job using a tooling must belong to the tooling information is mapped as follows by using formula (5):

[0220] f(production job, tooling information) = σ 生产作业.辅具编码＝辅具信息.辅具编码 (production job x tooling information).

[0221] The integrity constraint association of each production service unit abnormal alarm belonging to a production service unit is mapped as follows by using formula (5):

[0222] f(production service unit abnormal alarm, production service unit) = σ 生产服务单元..服务单元编码＝生产服务单元.服务单元编码 (production service unit abnormal alarm x production service unit).

[0223] The integrity constraint association of each production job corresponding to a production job is mapped as follows by using formula (5):

[0224] f(production job, production task) = σ执行报工.作业编码＝生产作业.作业编码 (Performing the report work x production work).

[0225] The integrity constraint relationship of each collected processing data corresponding to a production work is associated and mapped as follows by using formula (5):

[0226] f (processing data collection, production work) = σ 加工数据采集.作业编码＝生产作业.作业编码 (Processing data collection x production work).

[0227] The integrity constraint relationship of each collected data belonging to a manufacturing structure tree node is associated and mapped as follows by using formula (5):

[0228] f (processing data collection, manufacturing structure tree node) = σ 加工数据采集.制造码＝制造结构树节点.制造码 (Processing data collection x manufacturing structure tree node).

[0229] The integrity constraint relationship between each abnormal alarm record corresponding to the collected data is associated and mapped as follows by using formula (5):

[0230] f (abnormal alarm record, processing data collection) = σ 异常报警记录.记录编码＝加工数据采集.记录编码 (abnormal alarm record x processing data collection).

[0231] The integrity constraint relationship of each abnormal alarm record belonging to a manufacturing structure tree node is associated and mapped as follows by using formula (5):

[0232] f (abnormal alarm record, manufacturing structure tree node) = σ 异常报警记录.制造码＝制造结构树节点.制造码 (abnormal alarm record x manufacturing structure tree node).

[0233] The integrity constraint relationship of each quality inspection record belonging to a manufacturing structure tree node is associated and mapped as follows by using formula (5):

[0234] f (quality inspection record, manufacturing structure tree node) = σ 质检记录.制造码＝制造结构树节点.制造码 (quality inspection record x manufacturing structure tree node).

[0235] The integrity constraint relationship of each quality problem report corresponding to a quality inspection record is associated and mapped as follows by using formula (5):

[0236] f (quality problem report, quality inspection record) = σ 质量问题报告.记录编码＝质检记录.记录编码 (quality problem report x quality inspection record).

[0237] The integrity constraint relationship of each quality problem report belonging to a manufacturing structure tree node is associated and mapped as follows by using formula (5):

[0238] f (quality problem report, manufacturing structure tree node) = σ 质量问题报告.制造码＝制造结构树节点.制造码 (quality problem report x manufacturing structure tree node).

[0239] Given that the data generated during the manufacturing process of complex equipment all correspond to a node in the product manufacturing structure tree, and that the manufacturing structure tree can be generated by mapping the design structure tree, this invention proposes a mapping method between nodes in the manufacturing structure tree and nodes in the design structure tree, and an association mapping between data such as production processes, procedures, production tasks, task scheduling, production reporting, and quality inspection in the manufacturing stage and the manufacturing structure tree. By establishing data connection dependencies through data referential integrity mapping, the data in the manufacturing and design stages are linked.

[0240] Preferably, step S6: data association mapping during the operation and maintenance phase.

[0241] Preferably, step S6 includes steps S61-S62.

[0242] Preferably, step S61: Construct a node mapping between the product operation and maintenance structure tree and the design structure tree or manufacturing structure tree.

[0243] Preferably, step S611: establish a self-associative mapping of node data in the operation and maintenance structure tree.

[0244] The integrity constraint association mapping of the parent-child relationship of the product operation and maintenance structure tree nodes is applied using formula (5) as follows:

[0245] f(Operations and Maintenance Structure Tree Node, Operations and Maintenance Structure Tree Node) = σ 运维结构树节点.父节点编码＝运维结构树节点.节点编码 (Operations and Maintenance Structure Tree Nodes × Operations and Maintenance Structure Tree Nodes)

[0246] Preferably, step S612: establish the association mapping between the operation and maintenance structure tree nodes and the manufacturing structure tree nodes or design structure tree nodes.

[0247] The integrity constraint association mapping of product operation and maintenance structure tree nodes inherited from design structure tree nodes or manufacturing structure tree nodes is adopted using formula (5) as follows:

[0248] f(Operations and Maintenance Structure Tree Node, (Manufacturing Structure Tree Node ∪ Design Structure Tree Node)) =

[0249] σ 运维结构树节点.参照节点编码＝设计结构树节点.节点编码V运维结构树节点.参照节点编码＝制造结构树节点.制造码 (Manufacturing structure tree node × (Design structure tree node ∪ Manufacturing structure tree node), where: ∪ represents the union operation of the dataset, and V represents the conditional OR operation.

[0250] Preferably, step S62: establish a mapping between product operation and maintenance structure tree nodes and installation files, operation and maintenance records, perception data, fault data, and repair and replacement data during the operation and maintenance phase.

[0251] The integrity constraint association mapping for each equipment node file record belonging to an operation and maintenance structure tree node is as follows, using formula (5):

[0252] f (equipment node profile, operation and maintenance structure tree node) = σ 装备节点档案.节点编码＝运维结构树节点.节点编码 (equipment node profile x operation and maintenance structure tree node).

[0253] The integrity constraint association of each operation record belonging to an equipment node profile is mapped as follows by using formula (5):

[0254] f (operation record, equipment node profile) = σ 运行记录.档案编码＝装备节点档案.档案编码 (operation record x equipment node profile).

[0255] The integrity constraint association of each monitoring attribute belonging to an equipment node profile is mapped as follows by using formula (5):

[0256] f (monitoring attribute, operation and maintenance structure tree node) = σ 监控属性.节点编码＝运维结构树节点.节点编码 (monitoring attribute x operation and maintenance structure tree node).

[0257] The integrity constraint association of each collected perception data corresponding to a monitoring attribute is mapped as follows by using formula (5):

[0258] f (perception data, monitoring attribute) = σ 感知数据.属性编码＝监控属性.属性编码 (perception data x monitoring attribute).

[0259] The integrity constraint association of each collected abnormal data reflecting a monitoring attribute state is mapped as follows by using formula (5):

[0260] f (abnormal data, monitoring attribute) = σ 异常数据.属性编码＝监控属性.属性编码 (abnormal data x monitoring attribute).

[0261] The integrity constraint association of each collected abnormal data corresponding to an alarm information is mapped as follows by using formula (5):

[0262] f (abnormal data, alarm information) = σ 异常数据.告警编码＝告警信息.告警编码 (abnormal data x alarm information).

[0263] The integrity constraint association of each maintenance record corresponding to an operation and maintenance structure tree node is mapped as follows by using formula (5):

[0264] f (maintenance record, operation and maintenance structure tree node) = σ 检修记录.节点编码＝运维结构树节点.节点编码 (maintenance record x operation and maintenance structure tree node).

[0265] The integrity constraint association of each alarm information corresponding to a maintenance record is mapped as follows by using formula (5):

[0266] f (alarm information, maintenance record) = σ 告警信息.检修任务编码＝检修记录.检修任务编码 (alarm information x maintenance record).

[0267] The integrity constraint association mapping of each fault record corresponding to one operation and maintenance structure tree node by formula (5) is as follows:

[0268] f(fault record, operation and maintenance structure tree node) = σ 故障记录.节点编码＝运维结构树节点.节点编码 (fault record x operation and maintenance structure tree node).

[0269] The integrity constraint association mapping of each fault record corresponding to one maintenance record by formula (5) is as follows:

[0270] f(fault record, maintenance record) = σ 故障记录.检修任务编码＝检修记录.检修任务编码 (fault record x maintenance record).

[0271] The integrity constraint association mapping of each maintenance and replacement record corresponding to one operation and maintenance structure tree node by formula (5) is as follows:

[0272] f(maintenance and replacement record, operation and maintenance structure tree node) = σ 维修换件记录.节点编码＝运维结构树节点.节点编码 (maintenance and replacement record x operation and maintenance structure tree node).

[0273] The integrity constraint association mapping of each maintenance and replacement record corresponding to one maintenance record by formula (5) is as follows:

[0274] f(maintenance and replacement record, maintenance record) = σ 维修换件记录.检修任务编码＝检修记录.检修任务编码 (maintenance and replacement record x maintenance record).

[0275] In view that the data generated in the operation and maintenance process of complex equipment correspond to a node of the product operation and maintenance structure tree, and the node of the operation and maintenance structure tree can be obtained from the node of the manufacturing structure tree or the design structure tree, the application proposes the mapping relationship between the node of the operation and maintenance structure tree and the node of the design and manufacturing structure tree, and establishes the association mapping between the node of the operation and maintenance structure tree and the data such as the installation file, the operation and maintenance record, the sensing data, the fault alarm and the maintenance and replacement, and establishes the data connection dependency relationship through the data reference integrity mapping, so as to associate the data in the operation, manufacturing and design stages, and form the whole life cycle data system.

[0276] Step S7: multi-source heterogeneous data conversion and fusion storage method.

[0277] According to the steps of steps S3 to S6, the complex whole life cycle data warehouse mode of the complex equipment whole life cycle data fusion is established, and based on the data mode, the multi-source heterogeneous data is mapped and stored.

[0278] Preferably, step S7 includes steps S71 to S74.

[0279] Step S71: for the data object X, the data mode of the business system J is The data mode is composed of n attributes ​ its corresponding full life cycle data warehouse schema S x (A x1 ,A x2 ,…A xm ), is a data schema composed of m attributes A x1 ,A x2 ,…A xm , determines whether there is a reference attribute associated with the mapping established in A x1 ,A x2 ,…A xm :

[0280] If there is a reference attribute: check if the data object data it refers to has been imported. If not imported: remind to import the reference entity data object first and end; if imported: determine whether the reference attribute data set exists in all the reference entity attributes, and if not all exist, remind the user of the reference data anomaly;

[0281] Step S72: according to the similarity calculation mark in step S1, remove the attributes marked for deletion, and the remaining k attributes are matched according to the name or redundancy mark, and the mapping pairs of the k attributes in S x are matched

[0282] Step S73: for each attribute pair check and A xq data types and dimensions are consistent, and if not, convert.

[0283] Preferably, step S73 includes steps S731-S733.

[0284] Preferably, step S731: if the data types are not consistent, but are both numerical types (such as int and float, datetime and string, etc.), then call the unit conversion function of formula (6):

[0285]

[0286] In the formula: is the data type of , A xq .DataType is the data type of A xq , and this function calls the corresponding function in the type conversion library according to the data type of and the data type of A xq to realize conversion.

[0287] Preferably, step S732: if the dimensions are inconsistent (such as Km and m, or Fahrenheit and Celsius, etc.), then call the dimension conversion function of formula (7):

[0288]

[0289] wherein: is the data dimension of A xq .Dimension is A xq the data dimension of A, this function is according to the data dimension of A and A xq data dimension, calling the corresponding function of type conversion library to realize conversion.

[0290] Preferably, step S733: writing data into the data warehouse S x .

[0291] Preferably, step S74: repeating the process of S71-S73 for the other source of data object X business system data, for the data with the same main attribute value, only updating the null value attribute, and not repeating writing. x

[0292] In view of the data type conversion and data reduction problems in the integration of each heterogeneous data into a unified data mode in multi-source heterogeneous data fusion. In this regard, the application abstracts the data conversion type, establishes a multi-source heterogeneous data conversion and data reduction fusion algorithm, and solves the multi-source heterogeneous data extraction conversion problem.

[0293] The above examples are only used to illustrate the present application and not to limit the technical solutions described in the present application. Although the present application has been described in detail with reference to the above embodiments, the present application is not limited to the above specific embodiments, therefore any modification or equivalent replacement of the present application; and all technical solutions and improvements without departing from the spirit and scope of the application, which are all covered in the scope of claims of the present application.​

Claims

1. A method for constructing a multi-source heterogeneous data fusion mode for the entire lifecycle of complex equipment, characterized in that: It comprises the following steps: Step S1: identification and reduction of redundant attributes of multi-source heterogeneous data; Step S2: establishing a correlation mapping model between data entities; Step S3: product family meta-structure tree and knowledge class data correlation mapping; Step S4: design stage data correlation mapping; Step S5: manufacturing stage data correlation mapping; Step S6: operation and maintenance stage data correlation mapping; Step S7: multi-source heterogeneous data conversion and fusion storage method; Wherein, step S1 comprises step S11-step S12; Step S11: multi-source heterogeneous data main attribute marking; For the same data object , different methods of identifying and reducing redundant attributes are proposed for nominal attributes and numerical attributes respectively; first, the database dictionary is read to obtain the respective primary attributes PK( ) and PK( ), and it is judged whether PK( ) and PK( ) are globally unique identifiers generated by an algorithm or of the self-increment type, and the following 3 cases are handled: (1) If one of the two is a global unique identifier or a self-increment type, the attribute is marked for deletion, and the other attribute is marked as the main attribute; (2) If both are global unique or auto-increment type, remind the administrator to delete the two primary attributes from Select the main attributes in the remaining attribute set, and mark the two main attributes for deletion; (3) If both are neither global unique nor auto-increment type, judge whether the name and data type of both are same, if same, keep the mark of the one with more data samples as the primary attribute, if different, remind the administrator to select one from PK( ) and PK( ) as the primary attribute; ) Step S12: according to the attribute value range, the attributes are divided into nominal attributes and numerical attributes, and the attributes belonging to the same data type are reduced; step S12 comprises step S121-step S122; Step S121: nominal attribute reduction; Attribute A and attribute B are two attributes from different systems Attribute A has c values, A( ,…, ) and attribute B has r values B( ,…, ). Whether attribute A and attribute B describe the same feature is calculated as formula (1): = （1） wherein: observed frequency, expected frequency, The calculation formula (2) is as follows: = （2） ; the calculated If the correlation is true, compare the data type definition length of attribute A and B through the chi-square table, keep the attribute with the larger data type definition length, and mark the other attribute as redundant. Step S122: numerical attribute reduction; Attributes A and B are two attributes from different systems, and the correlation of attributes A and B is measured by calculating the covariance as shown in formula (3): - )( （3） wherein A and B are the mean values of the properties, n is the number of samples. The correlation coefficient of attributes A and B is further calculated from the covariance as shown in formula (4): （4） wherein: and respectively represent the standard deviation of the data corresponding to attributes A and B. Calculation results In comparison with the set threshold value, if the threshold value is exceeded: Compare the data type definition length of attributes A and B, and keep the attribute with longer data type definition length, and mark the other attribute as redundant; Wherein, step S2: establishing a correlation mapping model between data entities; Let and The data entities representing two different types of objects, the relationship data model to establish the two types of data entities between the reference integrity associated mapping as shown in formula (5): ) （5） wherein: and a mapping function, denotes a Cartesian product operation of two data sets, denotes a selection operation under the condition that the A attribute of the data pattern and the B attribute of the data pattern are equal, in the above mapping relationship, is called a referenced data relationship entity, , is called a referenced relationship entity, is the primary attribute set of . Wherein, steps S3-S6 establish correlation mapping according to the correlation mapping model of step S2.

2. The complex equipment whole life cycle multi-source heterogeneous data fusion mode construction method of claim 1, wherein: Step S3 comprises step S31-step S32; Step S31: constructing self-correlation mapping between product family meta-structure tree node data; Step S32: establishing mapping between product family meta-structure tree nodes and family model, rule knowledge data.

3. The complex equipment whole life cycle multi-source heterogeneous data fusion mode construction method of claim 2, wherein: Step S4 comprises step S41-step S42; Step S41: constructing product design structure tree and meta-structure tree node mapping; Step S412: establishing correlation mapping between design structure tree nodes and meta-structure tree nodes; Step S42: establishing mapping between design structure tree nodes and design model, design parameters, simulation analysis, strength analysis design stage data in the design stage.

4. The complex equipment whole life cycle multi-source heterogeneous data fusion mode construction method of claim 3, wherein: Step S5 comprises step S51-step S52; Step S51: constructing product manufacturing structure tree and design structure tree node mapping; Wherein, step S51 comprises step S511-step S512; Step S52: establishing mapping between manufacturing structure tree nodes and manufacturing stage process route, process, production task, task scheduling, production execution report, quality detection data.

5. The complex equipment whole life cycle multi-source heterogeneous data fusion mode construction method of claim 3, wherein: Step S6 comprises step S61-step S62; Step S61: constructing product operation and maintenance structure tree and design structure tree or manufacturing structure tree node mapping; Step S62: establishing mapping between product operation and maintenance structure tree nodes and operation and maintenance stage installation file, operation and maintenance record, perception data, fault data, maintenance and replacement data.

6. The complex equipment whole life cycle multi-source heterogeneous data fusion mode construction method of claim 5, wherein: Step S7 comprises step S71-step S74; Step S71: For data object X, the business system , is composed of n attributes The data pattern Its corresponding full lifecycle data warehouse model , consisting of m attributes The data pattern constitutes the judgment If there are any reference attributes established by the associated mapping, then process them as follows: If there is a reference attribute: check whether the referenced data object data has been imported; if not imported: remind to import the reference entity data object first and end; if imported: judge whether the reference attribute data set is in the referenced entity attribute, and if not all, remind the user of the abnormal reference data.

7. The complex equipment whole life cycle multi-source heterogeneous data fusion mode construction method of claim 6, characterized in that: Step S72: remove the attributes marked for deletion according to the similarity calculation in step S1, and the remaining k attributes Match by name or redundancy mark, match with The mapping pair of k attributes in , ),( , ),…,( , )}.

8. The complex equipment whole life cycle multi-source heterogeneous data fusion mode construction method of claim 7, characterized in that: Step S73: For each attribute pair ( , ),examine Check if the dimensions are consistent; if not, convert them. Step S73 includes steps S731-S733; Step S731: if the data types are inconsistent, but are both numerical types, then a unit conversion function of formula (6) is called: （6） To , To This function implements the conversion according to and data types by calling the corresponding functions in the type conversion library. Step S732: if the dimensions are inconsistent, then a dimension conversion function of formula (7) is called: （7） Step S733: writing data into the data warehouse .

9. The complex equipment whole life cycle multi-source heterogeneous data fusion mode construction method of claim 8, wherein: Step S74: Repeat the process of S71-S73 for other source of data object X's business system data, for data with same primary attribute value, only update the hollow value attribute, no duplicate write. Step S74: Repeat the process of S71-S73 for other source of data object X's business system data, for data with same primary attribute value, only update the hollow value attribute, no duplicate write.

Images