CN113487211A - Nuclear power equipment quality tracing method and system, computer equipment and medium - Google Patents
Nuclear power equipment quality tracing method and system, computer equipment and medium Download PDFInfo
- Publication number
- CN113487211A CN113487211A CN202110816567.6A CN202110816567A CN113487211A CN 113487211 A CN113487211 A CN 113487211A CN 202110816567 A CN202110816567 A CN 202110816567A CN 113487211 A CN113487211 A CN 113487211A
- Authority
- CN
- China
- Prior art keywords
- quality
- knowledge
- tracing
- nuclear power
- entity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 91
- 238000007499 fusion processing Methods 0.000 claims abstract description 12
- 230000014509 gene expression Effects 0.000 claims description 37
- 238000000605 extraction Methods 0.000 claims description 31
- 238000012545 processing Methods 0.000 claims description 30
- 238000004458 analytical method Methods 0.000 claims description 27
- 238000004519 manufacturing process Methods 0.000 claims description 15
- 230000004927 fusion Effects 0.000 claims description 9
- 238000013507 mapping Methods 0.000 claims description 9
- 238000003860 storage Methods 0.000 claims description 9
- 230000007547 defect Effects 0.000 claims description 8
- 239000000203 mixture Substances 0.000 claims description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 7
- 238000013461 design Methods 0.000 claims description 5
- 238000012800 visualization Methods 0.000 claims description 4
- 238000005065 mining Methods 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 claims description 3
- 238000012098 association analyses Methods 0.000 claims description 2
- 238000001228 spectrum Methods 0.000 claims description 2
- 238000010219 correlation analysis Methods 0.000 abstract description 4
- 239000013598 vector Substances 0.000 description 15
- 238000005516 engineering process Methods 0.000 description 14
- 238000003058 natural language processing Methods 0.000 description 11
- 230000011218 segmentation Effects 0.000 description 8
- 238000009434 installation Methods 0.000 description 7
- 238000004140 cleaning Methods 0.000 description 3
- 238000004590 computer program Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000011835 investigation Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 241001463014 Chazara briseis Species 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007596 consolidation process Methods 0.000 description 1
- 238000009430 construction management Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000010348 incorporation Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000011900 installation process Methods 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000000275 quality assurance Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 238000013024 troubleshooting Methods 0.000 description 1
- 238000011144 upstream manufacturing Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06395—Quality analysis or management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
- G06F16/353—Clustering; Classification into predefined classes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/268—Morphological analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
- G06F40/295—Named entity recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/022—Knowledge engineering; Knowledge acquisition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/027—Frames
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Economics (AREA)
- Computing Systems (AREA)
- Strategic Management (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Databases & Information Systems (AREA)
- Tourism & Hospitality (AREA)
- Educational Administration (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to a quality tracing method, a system, computer equipment and a medium for nuclear power equipment, wherein the method comprises the following steps: s1: acquiring related data for quality tracing; s2: constructing a knowledge graph for quality tracing; s3: and performing quality event tracing inquiry by using the knowledge graph for quality tracing. The nuclear power equipment quality tracing method provided by the invention can automatically extract knowledge from data resources related to nuclear power quality and perform fusion processing to form a knowledge graph for revealing the traceable association relation of quality characteristics, and update and expand the content of the knowledge graph through the knowledge. According to the input quality event, the related knowledge required by quality tracing stored in the knowledge map is utilized to drive the automatic tracing of the quality problem occurrence position along the quality forming path and the correlation analysis of the quality problem occurrence reason, so that the tracing of the quality event of the nuclear power equipment in the whole life cycle is realized.
Description
Technical Field
The invention relates to the field of nuclear engineering, in particular to a nuclear power equipment quality tracing method and system based on a knowledge graph, computer equipment and a medium.
Background
The nuclear power has the most strict requirements on quality safety, and a strictly established quality assurance and monitoring system is used for ensuring the safety of nuclear engineering. In the design, production, manufacture, installation, debugging and test operation processes of nuclear power equipment, if a quality event occurs, quality tracing with the quality event as an origin is a necessary link for ensuring the quality safety of nuclear power.
In the prior art, a quality tracing method of nuclear power equipment generally establishes an expert group, and carries out level-by-level investigation according to an enterprise established quality tracing process. The process has high dependence on expert experience, lacks of intelligent analysis means, is difficult to intercommunicate among platform resources in a nuclear power system, has long tracing flow and high tracing cost. Therefore, the nuclear power quality tracing related data resources are integrated and refined through an intelligent means, a structural information expression system for nuclear power equipment quality tracing is established, and an intelligent nuclear power equipment quality tracing method is necessary.
The knowledge graph is a technical method for describing the incidence relation between knowledge and all things in the world by using a graph model, and the knowledge graph technology is expected to effectively establish a structured expression system of quality tracing key information of nuclear power equipment, reflect key nodes and incidence relation in the quality forming process of the nuclear power equipment and further support the implementation of the quality tracing method.
However, since the knowledge graph technology has poor generality in different fields, in the process of applying the knowledge graph technology to the vertical field from the general field, the data source, the knowledge extraction target, the knowledge extraction method and the knowledge content structure of the knowledge graph need to be redesigned, and the implementation of the process needs professional background of the vertical field, professional knowledge of the knowledge graph technology and deep understanding of business. However, nuclear power equipment has complex technology and severe safety requirements, and how to select a data source for constructing a nuclear power equipment quality tracing knowledge graph, how to design a content structure for the nuclear power equipment quality tracing knowledge graph, and how to design a nuclear power equipment quality tracing method based on the knowledge graph become difficulties in hindering the application of the knowledge graph to nuclear power equipment quality tracing. Therefore, in the prior art, a knowledge graph is not used for quality tracing of nuclear power equipment, and an intelligent quality tracing method of the nuclear power equipment based on the knowledge graph is not provided.
Disclosure of Invention
The invention aims to provide a nuclear power equipment quality tracing method, a nuclear power equipment quality tracing system, computer equipment and a nuclear power equipment quality tracing medium based on a knowledge graph, wherein the knowledge graph is applied to information representation of nuclear power equipment quality tracing so as to realize rapid and intelligent tracing of nuclear power equipment quality events.
The invention provides a quality tracing method for nuclear power equipment, which comprises the following steps:
s1: acquiring related data for quality tracing;
s2: constructing a knowledge graph for quality tracing;
s3: and performing quality event tracing inquiry by using the knowledge graph for quality tracing.
Further, the related data for quality tracing includes structured data, semi-structured data, and unstructured data.
Further, the related data for quality tracing is derived from one or more of product quality defect data, product quality event related records, field standards and laws and regulations, quality plans, product supply relations and product production and manufacturing process files.
Further, step S2 further includes the steps of:
s21: and (3) knowledge extraction: extracting the relationship among the entity nodes, the entity attributes and the entities in the related data for quality tracing to form a standardized knowledge representation form and storing the standardized knowledge representation form;
s22: and (3) knowledge fusion: performing fusion processing on the knowledge obtained by the knowledge extraction, eliminating ambiguity in the knowledge, eliminating wrong and redundant knowledge, and meanwhile, counting the frequency/probability of the entity appearing in all event spaces, and storing the frequency/probability as the attribute of a corresponding node;
s23: knowledge processing: reasoning according to the knowledge obtained by the knowledge fusion to generate new knowledge, wherein the new knowledge comprises the relationship and the entity attribute between the entities obtained through reasoning;
s24: and (3) knowledge updating: and (4) iteratively updating the knowledge spectrum content by steps S21-S23 by using related data which are newly added in the later period and are used for quality tracing.
Further, in step S21, a natural language processing technique, a crawler technique and a batch script are used to perform knowledge extraction on the relevant data for quality tracing.
Further, the natural language processing technology comprises one or more of word segmentation, corpus processing, part of speech tagging, named entity recognition, entity relation extraction, attribute extraction and dependency syntactic analysis.
Further, the entity node in step S21 at least includes: quality problem expression nodes, quality problem occurrence object nodes, quality problem occurrence reason nodes, nuclear power equipment and nuclear power equipment sub-component nodes, and nuclear power equipment/sub-component related phase nodes and phase responsibility entity nodes; the entity attributes at least include: quality standard and normal state attribute of the equipment/component and probability attribute of the entity node in the total event space; the relationships between entities include at least: the relation between equipment/component/related stage and responsible party, the composition relation between equipment and equipment subcomponents, the association relation between quality problem expression and quality problem occurrence object, the relation between quality problem and cause causing quality problem occurrence, and the circulation relation of equipment/component manufacturing and installing flow.
Further, the method of knowledge fusion in step S22 includes one or more of entity disambiguation, coreference resolution, and knowledge merging.
Further, the method for counting the frequency/probability of occurrence of the entity in the whole event space in step S22 includes: by counting the quality problems recorded in different quality events, the probability/frequency of a certain entity appearing in all event spaces is obtained based on the statistical principle.
Further, in step S23, new knowledge is generated using logic-based reasoning and/or graph-based reasoning.
Further, step S3 further includes the steps of:
s31: intelligently analyzing the input quality event, and identifying the entity, entity relationship and quality problem expression information of the input quality event;
s32: mapping the information identified in S31 to nodes and relationships in the knowledge-graph;
s33: and driving search based on the nodes, attributes and relationship knowledge stored in the knowledge graph, and tracing the nodes and reason information generated by the input quality problem.
Further, the method for performing intelligent analysis in step S31 includes one or more of word segmentation, part-of-speech tagging, named entity recognition, and entity relationship extraction.
Further, the search in step S33 is: the node, the attribute and the relation corresponding to the state expression of the quality problem are positioned in the quality tracing knowledge graph, the node, the attribute and the relation stored in the knowledge graph are utilized to carry out reverse tracing on the object with the quality problem along a quality forming path of nuclear power equipment, and the generation position, the state expression and the reason of the quality problem and the normal state information of the equipment are compared step by step and associated analysis is carried out according to the generation position, the state expression and the reason of the quality problem and the normal state information of the equipment, so that the reasoning result of the reason with the quality problem is given, the reliability is calculated according to the probability weight of the node, the stage to which the quality problem belongs is judged, and a tracing chain of the stages with quality defects, phenomena, objects and related is formed.
Further, the quality problem stage comprises the stages of design, purchase, construction and debugging of nuclear power equipment.
The invention provides a nuclear power equipment quality tracing system on the other hand, which comprises:
the data resource acquisition module is used for acquiring data resource information required by quality tracing from various data sources;
the knowledge map construction module is used for processing various data acquired by the data resource acquisition module and extracting knowledge information in the data;
and the quality tracing algorithm module is used for searching the knowledge graph according to the input quality event drive, inquiring the quality problem generation node and possible reasons of the nuclear power equipment according to the position and quality state of the quality problem, and performing visual display.
Further, the data resource acquisition module comprises:
the data acquisition unit is used for acquiring related data resources for quality tracing from a website, an enterprise KMS system and a platform database;
and the data index unit is used for establishing an index for the data resources obtained by the data acquisition unit and classifying the data resources.
Further, the knowledge graph building module comprises:
the knowledge extraction unit is used for processing the data resources acquired by the data resource acquisition module and extracting the entity, entity attribute and entity relation knowledge in the data resources;
the knowledge fusion unit is used for carrying out disambiguation and fusion processing on the multi-source knowledge so as to improve the knowledge quality;
the knowledge processing unit is used for carrying out deep mining based on the existing knowledge and improving the breadth and depth of the knowledge;
and the knowledge updating unit is used for updating and expanding the existing knowledge graph.
Further, the quality tracing algorithm module comprises:
the intelligent analysis unit is used for extracting quality event related entity information from the input quality events to be inquired and mapping the quality event related entity information to the corresponding position of the quality tracing knowledge graph;
the quality tracing indexing unit is used for tracing quality problems, tracing from the node positioned by the intelligent analysis unit in the quality tracing knowledge graph along the entity relationship, comparing the node information recorded by the graph with the information of the input quality event, and finding out a responsible entity causing the quality event;
the tracing process visualization unit is used for visually displaying the quality problem tracing process and the quality problem propagation path;
and the analysis report generating unit is used for generating a quality tracing result analysis report according to the established enterprise template.
Yet another aspect of the present invention provides a computer apparatus comprising: a processor and a memory for storing processor-executable instructions; the processor is configured as a method as may be described above.
Yet another aspect of the present invention provides a computer-readable storage medium having stored therein computer-executable instructions that, when executed by a processor, cause the processor to perform the steps of the method as described above.
The nuclear power equipment quality tracing method provided by the invention can automatically extract knowledge from data resources related to nuclear power quality and perform fusion processing to form a knowledge graph for revealing the traceable association relation of quality characteristics, and update and expand the content of the knowledge graph through the knowledge. According to the input quality event, the related knowledge required by quality tracing stored in the knowledge map is utilized to drive the automatic tracing of the quality problem occurrence position along the quality forming path and the correlation analysis of the quality problem occurrence reason, so that the tracing of the quality event of the nuclear power equipment in the whole life cycle is realized. The system breaks through the traditional time-consuming and labor-consuming manual investigation method in the nuclear power quality problem tracing process, comprehensively improves the quality tracing efficiency of nuclear power equipment, greatly reduces the cost and the personnel dependence, and realizes the effective and reliable guarantee technology of safety, quality and environment of nuclear power.
Drawings
FIG. 1 is a schematic flow chart of a nuclear power equipment quality tracing method based on a knowledge graph according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of construction of a knowledge graph in the nuclear power equipment quality tracing method based on the knowledge graph provided by the embodiment of the invention;
FIG. 3 is a schematic flow chart illustrating a query process using a knowledge graph in the nuclear power equipment quality tracing method based on the knowledge graph according to the embodiment of the present invention;
fig. 4 is a schematic structural diagram of a nuclear power equipment quality tracing system based on a knowledge graph according to another embodiment of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
As shown in fig. 1, an embodiment of the present invention provides a nuclear power equipment quality tracing method based on a knowledge graph, including the following steps:
s1: and acquiring related data for quality tracing.
The related data for quality tracing can be acquired from a nuclear power intelligent construction platform (such as NICE) platform, a construction management platform, an Internet of things platform and other systems inside a nuclear power company, and specific platform systems may be different with different nuclear power companies. The related data comprises structured data, semi-structured data and unstructured data. Specifically, the structured data is data in a fixed format, such as management information, such as provider system information and product supply relationship, stored in a relational database; the semi-structured data is data of a non-relational model and has a basic fixed structure mode, such as log files in an internal engineering management website of a nuclear power company, site quality witness records in a picture format and the like; the unstructured data is data which cannot exist in a fixed format, such as quality event lists, experience feedback lists, field standard documents and the like recorded in a text description mode in the manufacturing process of nuclear power equipment.
The related data for quality tracing is derived from one or more of product quality defect data, product quality event related records, field standards and laws and regulations, quality plans, product supply relations and product production and manufacturing process files, and can reflect the quality forming process of nuclear power equipment.
Specifically, the quality forming process of the nuclear power equipment comprises the following steps: the series connection of key node information related to quality in the process of design-purchase-construction-debugging stage of nuclear power equipment, for example, component information of certain nuclear power equipment, quality event lists of certain nuclear power equipment and subcomponents thereof at different stages, generated experience feedback and the like, can reflect a phenomenon-object-related stage link generated by quality defects of the nuclear power equipment and an incidence relation between objects, and further can reflect responsibility entities corresponding to the objects and the related stages and other information beneficial to improving the accuracy and content richness of quality tracing results. The specific selection of the key nodes can be confirmed according to the characteristics of the nuclear power equipment and actual tracing requirements, but at least comprises the composition information of the nuclear power equipment, the design-manufacture-installation-debugging process information of the nuclear power equipment and the subcomponents thereof, the responsibility party information of each link, the quality index information of the nuclear power equipment and the subcomponents thereof, and the quality problem expression and reason information of the nuclear power equipment.
S2: and constructing a knowledge graph for quality tracing.
As shown in fig. 2, step S2 further includes the following steps:
s21: and (3) knowledge extraction: processing unstructured data in the relevant data for quality tracing using natural language processing techniques, wherein the natural language processing techniques comprise: using one or more of word segmentation, corpus processing, part of speech tagging, named entity identification, entity relationship extraction, attribute extraction and dependency syntactic analysis; processing semi-structured data in the related data for quality tracing by using a crawler technology; and carrying out standardized conversion on the structured data in the related data for quality tracing by using a batch script. Through the processing, the entity nodes, the entity attributes and the relationships among the entities in the data are extracted to form a standardized knowledge expression form, and the knowledge expression form is stored by using a graph database and a relational database. Wherein, the entity node includes at least: the system comprises quality problem expression nodes, quality problem generation object nodes, quality problem generation reason nodes, nuclear power equipment and nuclear power equipment sub-component nodes, and nuclear power equipment/sub-components relate to phase nodes and phase responsibility entity nodes. The entity attributes at least include: quality standard and normal state attributes of the device/component, and probability attributes of the node occurring in the total event space. The relationships between entities include at least: the relation between equipment/component/related stage and responsible party, the composition relation between equipment and equipment subcomponents, the incidence relation between quality problem expression and quality problem occurrence object, the relation between quality problem and cause causing quality problem occurrence, and the circulation relation of equipment/component manufacturing and installing flow.
Specifically, data cleaning is required before processing the related data, and the method includes: missing values in the text data are supplemented or abandoned, duplicate data are removed, operations such as word removal and wrongly written character detection are carried out on the text information, the quality of original data can be improved through data cleaning, and therefore the accuracy of information extraction, the efficiency of a processing process and the accuracy of a tracing result are improved. The missing value refers to a situation that some data in some samples is missing in a batch of data samples or some characteristic value is not recorded, and the reasons for generating the missing value may be manual recording negligence, storage medium failure and sampling omission. For example, the failure behavior of the device in the quality incident ticket is not filled in, or a parameter of an individual part in parts manufactured in the same batch is not recorded. For parts produced in the same batch, parameter values are distributed uniformly, and under the condition that other sample data in the same batch are complete, the missing values can be completed by taking the average number and the mode according to the distribution rule of the data. For missing values that cannot be filled, the sample is discarded and not extracted.
It can be understood that, for unstructured text information, one or more of word segmentation, corpus processing, part of speech tagging, named entity identification, entity relation extraction, attribute extraction and dependency syntactic analysis can be used for extracting key information from raw data subjected to data cleaning, and key node information recorded in the text information, such as phenomenon-object-related stage generated by quality defect of nuclear power equipment, and attributes and association relations of the key node information are extracted. For example, the BERT model may be used to perform word segmentation, named entity recognition, and the like on text information. For the problem of recognition of proper nouns in some nuclear power vertical fields, a nuclear power equipment quality related corpus is used for constructing a nuclear power equipment quality tracing field corpus, so that the recognition precision of the proper vocabularies in the nuclear power vertical fields can be improved, wherein the larger the magnitude of the corpus is, the better the magnitude of the corpus is, and generally, the number of the corpus is not less than thousands of related sentences. Firstly, on the basis of open-source natural language processing tools, such as NLTK, HANLP and the like, raw linguistic data are processed, and for vertical fields such as nuclear power, due to the fact that the expression mode, proper nouns and short names of the open-source natural language processing tools have particularity, the open-source natural language processing tools have poor effect on processing the linguistic data in the nuclear power field and need to be further corrected on the basis; therefore, the processing result needs to be corrected according to expert experience and by combining with the quality tracing practical application scene of the nuclear power equipment, so that the idiom is formed, and a corpus training model is used for processing more data. Such as: for a field quality event recorded in a quality event sheet: and leakage of steam from the flange of the main steam filter. The reason is as follows: (1) on-site inspection shows that the flange bolt torque of the main steam filter is insufficient, so that steam leakage occurs; (2) the steam filter flange gasket is a tooth-shaped gasket which has higher installation requirements, and the gasket and the installation process are easy to cause steam leakage due to slight deviation, obviously, the text information is unstructured data, and a data processing method of the unstructured data is used for the text information. "main steam filter", "main steam filter flange", "flange bolt", "flange gasket", "profile of tooth gasket", the quality problem who discerns shows: "steam leakage", the identified quality problem occurrence objects are: the "primary steam filter flange", the identified cause is: "insufficient moment" and "mounting deviation", the identified relationship is: "main steam filter" and "main steam filter flange", "main steam filter flange" and "flange bolt", "main steam filter flange" and "flange gasket" belong to the composition relation between equipment and equipment subcomponents, "steam leakage" and "main steam filter flange" belong to the incidence relation between quality problem performance and quality problem occurrence object, "steam leakage" and "moment is insufficient", "installation deviation" belong to the relation between quality problem and cause that quality problem occurs, the identified attribute is: the flange gasket is a tooth-shaped gasket. The above process logic may be interpreted as: the method includes the steps that a natural language processing tool is needed to process texts in the quality tracing field of nuclear power equipment- > the natural language processing tool in the general field cannot obtain good effects in the nuclear power field- > a part of texts in the nuclear power field are taken out firstly, the natural language processing tool in the general field is used for processing, processing results with low precision- > are obtained and corrected based on the processing results, high-precision cooked corpora are formed- > a corpus in the nuclear power quality tracing field is constructed by using the cooked corpora, and the BERT model is trained by using the corpus- > the trained model is used for processing the texts in all the quality tracing fields of the rest nuclear power equipment.
It can be understood that for semi-structured data, a crawler technology can be used, and the key information in the semi-structured data can be extracted from a data source such as an internal engineering website of a nuclear power company by means of regular matching and the like. Such as: the quality management information platform of the nuclear power company stores a digitized historical quality event list, the platform can be accessed from an internal website of the nuclear power company, and the quality event list is presented by a plurality of pieces of data on a webpage, such as: the method comprises the following steps of event list numbering, time, position, related stage, quality event content and the like, wherein each piece of data has a label, but the forms of the data are different, for example, information simultaneously existing on a website page can be as follows: unstructured picture information, quality problem text record information, quality problem reason troubleshooting text record information, structured time, equipment numbers, unit numbers and the like, contents under a target label can be extracted by utilizing a crawler technology, and a data set only containing target information is formed.
It is understood that for the structured data, a batch script and some open source tools (such as RDB2RDF mapping languages: dm (direct mapping) and R2RML) can be used to extract a useful part (i.e. a part related to the aforementioned target extraction content) in the structured data, and corresponding information is extracted from a data source such as an excel table or a relational database of enterprise record data according to an index. Such as: in order to obtain the standardized knowledge expression of quality problem expression, reason and object information which is arranged and stored in a relational database by an enterprise, target fields in the database are extracted by using a batch script according to labels, information is converted into RDF (resource description framework) graph data from the relational database by using a DM (direct data format) direct mapping language, and the structure of the relational database is directly reflected in an RDF graph.
Specifically, when extracting entity relationships, a predefined relationship type is usually required, and in a specific embodiment of the present invention, the predefined relationship type at least includes: the relation between equipment/component/related stage and responsible party, the composition relation between equipment and equipment subcomponents, the incidence relation between quality problem expression and quality problem occurrence object, the relation between quality problem and cause causing quality problem occurrence, and the circulation relation of equipment/component manufacturing and installing flow. The representation of these relationships in the graph database may be: for example, the relationship between the quality problem representation and the quality problem occurrence object is as follows: (object) - [: Performance ] - > (Performance), the relationship between the equipment sub-components and the equipment is: (device) - [: by the composition ConsistOf ] - > (sub-component), another device quality problem performance caused by a quality problem of a certain device can be expressed as: (device a) - [: Performance ] - > (device a Performance) - [: result in ResultIn ] - > (device B) - [: Performance ] - > (device B Performance).
It should be noted that the relationship types predefined in the embodiment of the present invention are proposed in the context of quality tracing in the nuclear power field, have uniqueness and creativity, and have a heuristic effect on those skilled in the art, and those skilled in the art may also define more relationship types (which cannot be exhaustive) in combination with actual requirements and data characteristics on this basis to meet business needs.
Preferably, the result of the knowledge extraction at least comprises: the method comprises the steps of nuclear power equipment manufacturing process link nodes and input and output relations thereof, system or equipment entities and subsystem or subcomponent entities and relations thereof, system or equipment normal state information, historical quality problem expression, quality problem corresponding objects, quality problem related stages, quality problem generation reasons and relations between the quality problems and the quality problem generation reasons.
Specifically, the knowledge extraction result is subjected to knowledge standardized representation and stored, the standardized knowledge representation form can be an RDF triple form or an attribute graph form, and the extracted standardized knowledge information can be stored by using a graph database Neo4j and a relational database MySQL. For example, a D2R tool may be used to map a relational database into an RDF format, or a knowledge expression in the form of an attribute graph may be established in the form of a CREATE of Neo4j, and it is understood that a nuclear power equipment quality traceability knowledge graph stored in the form of an attribute graph may include the following main structures: (1) and (3) node: the quality problem representation node, the quality problem occurrence object node, the quality problem occurrence reason node, the nuclear power equipment/sub-component related phase node, the nuclear power equipment component node and the responsible entity node are included, wherein the quality problem occurrence object node and the nuclear power equipment/nuclear power equipment sub-component node can be overlapped, and when the quality problem occurrence object node and the nuclear power equipment/nuclear power equipment sub-component node are overlapped, the node has two node labels; (2) the relationship is as follows: the relation between equipment/component/related stage and responsible party, the composition relation between equipment and equipment subcomponents, the incidence relation between quality problem expression and quality problem occurrence object, the relation between quality problem and cause causing quality problem occurrence, and the equipment/component manufacturing and installing process circulation relation; (3) the attributes are as follows: quality standards and normal state attributes of the devices/components, probability attributes of the occurrence of nodes in the total event space, and supplementary descriptions of nodes or relationships, etc. For example, based on the knowledge extraction result of the field quality event recorded in a certain quality event list, a creating method of Neo4j is used to CREATE a knowledge expression in the form of an attribute graph, and the specific steps may be: (1) according to the extracted named entity, the identified quality problem expression, the identified quality problem generation object and other knowledge information, creating each node in a graph database by using a CREATE statement in a Neo4j database operation language Cypher, defining a node label of an Equipment node as Equipment, a node label of a quality problem expression node as Performance, and a quality problem generation Reason node label as Reason, for example: the quality problem expression node with the content of ' steam leakage ' can be created in the graph database Neo4j by inputting CREATE (Performance { Performance: ' steam leakage) }. The creation statement of the node is structured, and a new creation statement can be generated only by replacing the node label and the node content in the statement; (2) according to the extracted relationship, an association relationship is created for the newly created nodes in the graph database, such as: inputting MATCH (a: Performance { Performance: "steam leakage" }), (b: Equipment { efficiency: "main steam filter flange" }), matching a node a with the content of steam leakage and a node b with the content of main steam filter flange, and then using CREATE (a) - [: Performance ] - > (b) to create an incidence relation between the quality problem expression and the quality problem generation object between the nodes a and b; (3) creating attributes for the newly created nodes in the graph database according to the extracted attributes; (4) and screening the nodes which finish the relation connection and attribute addition, and deleting the independent nodes which are not in contact with other nodes.
Specifically, the original knowledge graph obtained by extracting knowledge from the original data may contain redundant and erroneous information, and at this time, the knowledge needs to be fused to ensure the quality of the knowledge.
S22: and (3) knowledge fusion: and performing fusion processing on the knowledge obtained by the knowledge extraction, eliminating ambiguity in the knowledge, eliminating wrong and redundant knowledge in the knowledge to ensure the quality of the knowledge, and meanwhile, counting the frequency/probability of the entity appearing in all event spaces and storing the frequency/probability as the attribute of the corresponding node.
Preferably, the method for fusion processing of knowledge comprises: entity disambiguation, coreference resolution, knowledge consolidation, or a combination thereof. The entity disambiguation technique can be referred to as: [1] BaggaA, Baldwin b, entity-Based Cross-Document center Using the vector.2002.[2] Pedersen T, purandaree a, Kulkarni a. name correlation by Clustering Similar Contexts [ C ]// Computational threads and intellectual Text Processing,6th International Conference, CICLing 2005, Mexico City, Mexico, February 13-19,2005, proceedings. The coreference resolution technology can refer to: [1] riloff E M, Bean D.upstream learning of connected roller Knowledge for collaborative analysis [ J ]. procofhlt/naac, 2004 [2] Cheng T, LauW H W, Paparidios S.Entitution synthases for Structured Web Search [ J ]. IEEE Transactions on Knowledge & Data Engineering,2012,24(10): 1862-membered 1875 ]; the knowledge incorporation techniques can be referred to as: [1] michel F, Montagnat J, Faron-Zucker C.A surfey of RDB to RDF translation approaches and tools [ J ]. base de donn es, 2014; and will not be described in detail herein.
Preferably, the method for counting the frequency/probability of occurrence of the entity in the whole event space comprises: by counting the quality problems recorded in different quality events, the probability/frequency of an entity appearing in the total event space is obtained based on the statistical principle and is stored as an attribute of a corresponding node or edge, and it can be understood that entities mentioned many times in a single event record are counted only once.
It can be understood that entity disambiguation is a technique for solving the ambiguity of homonymic entities, for example, word vectors can be constructed in combination with semantic features, cosine similarity is calculated using the word vectors, whether the nominated words and homonymic entities in the knowledge graph represent different meanings is judged, and the nominated items are clustered to corresponding entities; it is to be understood that coreference resolution is a technique for resolving the problem of different referents pointing to the same entity, e.g., in the same mass event report sheet, the terms "condensate pump", "water pump", "pump" may be used to refer to the same pump body, for example, for the deep learning method, a circulation neural network is built by using an LSTM unit based on an expression sequencing model, so that the expressed global characteristics can be obtained to improve the accuracy of a coreference resolution task, and different reference words are linked to the same entity; it can be understood that knowledge merging is a method for acquiring knowledge input from an external knowledge database, for example, "condensate pump", "water pump", "pump" are the same pump body, or "pump" appearing in different scenes refers to different entities, it is obvious that there is ambiguity of the reference between different knowledge bases, and when fusing a plurality of knowledge bases, merging operation needs to be performed on the references; for example, the data layers of the two knowledge bases are fused, the names, attributes and relationships of the entities are combined, or some open-source mapping tools such as XSPARQL and the like are used to convert the knowledge information stored in the structured data/semi-structured data into a standard format available for a knowledge graph such as RDF in batch. Furthermore, in the knowledge fusion process, the probability of the same entity occurring can be calculated based on the statistical principle: by counting the quality problems recorded in different event records, the probability/frequency of occurrence of an entity in the total event space is obtained and stored as an attribute of a corresponding node or edge, and it can be understood that entities mentioned many times in a single event record are counted only once. For example, for a pump of the same type, assuming that different quality problems are recorded in different event lists, before performing fusion processing on an entity of the pump of the type, statistics may be performed on the probability of occurrence of fault expression in the event list related to the pump, and the probability may be stored as an attribute in a corresponding quality problem expression node, and when performing inference query based on a knowledge graph subsequently, the probability attribute may be used as a decision reference.
S23: knowledge processing: and reasoning by using a knowledge reasoning technology according to the existing knowledge in the database to generate new knowledge. Wherein the new knowledge comprises: and obtaining the relationship and entity attribute between the entities through reasoning.
Preferably, the knowledge inference technique comprises: one or more of logic-based reasoning and graph-based reasoning.
Specifically, the knowledge inference is to perform inference according to existing entity, relationship, and attribute information in a knowledge graph to obtain a new relationship between entities, for example, knowledge recorded in an experience feedback sheet reflects that a device using a certain sub-component may generate quality problems when operating in a certain environment, and it can be understood that other devices using the sub-component may also have a probability to generate quality problems when operating in the environment, so that a new relationship may be added between other devices using the sub-component and the quality problems, and a probability parameter is given, which may be calculated from a similarity between two device object entities and a probability attribute of an object. For example, logic-based reasoning can use the Tableux method, and open source reasoning tools such as Hermit and the like are applied; graph-based reasoning can use the Path ranking algorithm to perform reasoning based on nodes and relationships stored in a graph database, which can be interpreted as: starting from an initial node, randomly walking according to nodes and relations in a graph database, taking a generated path as a characteristic, training a Logistic Regression classifier by utilizing the characteristic, and applying the characteristic to relational reasoning.
S24: and (3) knowledge updating: and iteratively updating the content of the knowledge graph. And processing the related data of the newly added quality tracing through the steps, supplementing the knowledge map, and realizing incremental updating of knowledge.
S3: and performing quality event tracing query by using the knowledge graph.
As shown in fig. 3, step S3 further includes the following steps:
s31: and intelligently analyzing the input quality event, and identifying key information such as entities, fault expression and the like.
Preferably, the intelligent analysis method comprises: word segmentation, part of speech tagging, named entity recognition, entity relationship extraction and dependency syntactic analysis. The intelligent analysis result is the quality event related entity, entity relation and quality problem expression information.
Specifically, through intelligent analysis of the field quality event list, key information extraction can be performed on the quality event list by using one or more of word segmentation, corpus processing, part of speech tagging, named entity identification, entity relationship extraction, attribute extraction and dependency syntactic analysis, and primary phenomena and primary objects of the nuclear power equipment quality events recorded by text information, as well as attributes and incidence relations of the primary phenomena and the primary objects, can be extracted. For example, the BERT model may be used to perform word segmentation, named entity recognition, and the like on text information. For the problem of recognition of proper nouns in some nuclear power vertical fields, a nuclear power equipment quality related corpus is used for constructing a nuclear power equipment quality tracing field corpus, and the recognition precision of the proper vocabularies in the nuclear power vertical fields can be improved. It will be appreciated that the extracted information is likewise converted into a standard representation of the knowledge-graph.
S32: and mapping the identified information to nodes and relations in the knowledge graph.
Specifically, the primary phenomena and the primary objects of the quality events of the nuclear power equipment extracted from the input quality events, and the attributes and the association relations of the primary phenomena and the primary objects are converted into database query statements, the database query statements are directly matched in the knowledge graph, for example, a Match item is directly searched in Neo4j by using a Match statement provided by a database query language, screening is performed according to the contents, the attributes and the association relations of the nodes identified in the input information, and the searched result is the matching object of the input information in the knowledge graph. Or matching the query object to the corresponding node, relation and attribute in the knowledge graph through semantic similarity analysis and other modes. Specifically, for example, based on a word vector space formed after words are stopped by text information, word vectors are constructed by using a word2vec model, the word2vec model is an NLP model proposed by Google, all words can be represented as low-dimensional dense vectors, context relations are stored in calculated word vectors, and therefore the same words in different contexts can be distinguished, for example, different types of water pumps running in different scenes of different units are recorded in two quality event lists, and because information contained in the contexts of the word vectors is different, the word vectors calculated by the word2vec are different and cannot be recognized as the same entity. The cosine similarity is calculated by using the word vector, and the similarity between words can be qualitatively measured. Specifically, one calculation formula of the cosine similarity may be:
in the formula, x and y are word vectors of two vocabularies. It can be understood that, for any stored node or relation in the knowledge graph, the self, the forward node, the backward node and the incidence relation of the node or relation form a vector space, word vectors of the node or relation can be respectively calculated, cosine similarity between the input information word vectors and the stored node or relation word vectors of the knowledge graph is calculated, and the input information is clustered to the corresponding node or relation in the knowledge graph according to the cosine similarity.
A database query statement is a computer program for querying data from a database, written in a database operating language, such as the Cypher language used by Neo4j database databases. For the problem of how to convert the query into the database query language, because the operation language of the database is a structured language, it can be understood that a query statement program can be designed in advance, and when the query is needed, the information such as the node to be queried, the relationship and the like is used for replacing the corresponding field in the original query statement to generate a new query statement.
S33: and (4) based on knowledge-driven searching of nodes, attributes, relations and the like stored in the knowledge graph, tracing the nodes and reason information generated by the quality problem.
Preferably, the knowledge graph-based search is to locate nodes, attributes and relations corresponding to the state expression of the quality problem in the quality tracing knowledge graph, reversely trace the object with the quality problem along the quality forming path of the nuclear power equipment by using the nodes, attributes and relations stored in the knowledge graph, perform step-by-step comparison and association analysis according to the occurrence position of the relevant historical quality problem, the state expression and reasons of the quality problem and the normal state information of the equipment, give an inference result of the cause of the quality problem, calculate the reliability according to the probability weights of the nodes and the edges, judge the stage to which the quality problem belongs, and form a tracing chain of the stages of quality defect generation-phenomenon-object-related. Wherein the stages comprise: designing, purchasing, constructing and debugging nuclear power equipment.
Specifically, traversal query is carried out along associated facts in the knowledge graph according to nodes, relations and attributes positioned in the knowledge graph, whether the node state of each device is different from the normal state of the device or not is continuously compared, the probability of each path is calculated according to parameters such as probability attributes, and finally quality tracing query results based on the knowledge graph are given according to the occurrence probability sequence. For example, suppose that a quality event of flange steam leakage of a main steam filter occurs in a debugging site, and only the steam leakage phenomenon of the main steam filter is recorded in an input site quality event list, but the position and the reason of the steam leakage are not specifically checked. By applying the nuclear power equipment quality tracing method based on the knowledge graph provided by the invention, the relevant initial nodes, attributes and relations are matched in the knowledge graph according to the nodes, the relations and the attribute information extracted from the input information, the nodes matched into the knowledge graph by the input information are 'main steam filter' and 'steam leakage', and the traversing search is carried out from the initial nodes along the relation path according to the existing nodes and relation graphs in the knowledge graph: firstly, a main steam filter flange and a main steam filter flange are arranged at two nodes of a main steam filter and a steam leakage node, the intermediate node of the main steam filter and the main steam filter flange are sub-components of the main steam filter, so that a specific component causing the steam leakage of a device node of the main steam filter is presumed to be the main steam filter flange, the steam leakage node can reach a node with insufficient torque and a node with installation deviation along the relationship between quality problems and causes of quality problems, the node with insufficient torque is traced back to the flange bolt node, the node with installation deviation is traced back to the flange gasket node, and the flange bolt and the flange gasket belong to the sub-components of the main steam filter flange according to a knowledge graph, so that the traversal of all paths is completed, and the probability of each path is calculated according to the probability attributes of the nodes, assuming that the number of times of occurrence of "insufficient torque" is more than the number of times of occurrence of "mounting deviation" in the total event space, returning query results in the order of occurrence probability from high to low as: the most likely location for causing steam leakage from the main steam filter is the "main steam filter flange", for possible reasons: (1) insufficient flange bolt torque, (2) flange gasket installation deviation; in another embodiment of the present invention, for providing the specific state information of the equipment, the values of these attributes may be compared according to the normal state or quality standard attribute of the node stored in the knowledge graph, so as to find an abnormal node or exclude a normal node, for example, the normal state of the flange bolt stored in the quality tracing knowledge graph of the nuclear power equipment is that the torque is greater than or equal to the value a, and the engineer records that the measured torque of the flange bolt of the main steam filter is the value B in the field quality event list, and it can be understood that in the extraction result of the input information this time, the "flange bolt" node includes the attribute "torque" B ". In the derivation process based on the knowledge graph, when a flange bolt node is reached, the attribute 'moment ≥ A' of the flange bolt node is compared with the attribute 'moment ≥ B' of the flange bolt node in input information, if B ≥ A, the flange bolt node is judged not to be the position where the quality problem of steam leakage of the main steam filter occurs, and the 'insufficient moment' connected with the flange bolt node is not the reason of steam leakage, so the final tracing result is as follows: the most likely location for causing steam leakage from the main steam filter is the "main steam filter flange", for possible reasons: flange gasket installation tolerances.
The nuclear power equipment quality tracing method provided by the invention can automatically extract knowledge from data resources related to nuclear power quality and perform fusion processing to form a knowledge graph for revealing the traceable association relation of quality characteristics, and update and expand the content of the knowledge graph through the knowledge. According to the input quality event, the related knowledge required by quality tracing stored in the knowledge map is utilized to drive the automatic tracing of the quality problem occurrence position along the quality forming path and the correlation analysis of the quality problem occurrence reason, so that the tracing of the quality event of the nuclear power equipment in the whole life cycle is realized. The system breaks through the traditional manual checking method in the nuclear power quality problem tracing process, comprehensively improves the nuclear power equipment quality tracing efficiency, greatly reduces the cost and personnel dependence, and realizes the effective and reliable guarantee technology of nuclear power safety, quality and environment (namely 'safety ring')
As shown in fig. 4, another embodiment of the present invention provides an architecture diagram of a quality tracing system based on a knowledge graph, including:
a data resource obtaining module 100, configured to obtain data resource information required for quality tracing from various data sources;
the knowledge map construction module 200 is used for processing various data acquired by the data resource acquisition module and extracting knowledge information in the data;
the quality tracing algorithm module 300 is used for searching the knowledge graph according to the input quality event drive, finding the quality problem generation node and possible reasons of the nuclear power equipment according to the position and quality state of the quality problem, and performing visual display.
In a specific embodiment, preferably, the data resource acquiring module 100 includes a data acquiring unit 101 and a data indexing unit 102. The data acquisition unit is used for acquiring related data resources for quality tracing from a website, an enterprise KMS system and a platform database, and the data indexing unit is used for establishing indexes for the data resources acquired by the data acquisition unit and classifying the data resources.
In one embodiment, the knowledge graph building module 200 includes a knowledge extraction unit 201, a knowledge extraction unit 202, a knowledge processing unit 203, and a knowledge update unit 204. The knowledge extraction unit 201 is configured to process the data resources obtained by the data resource obtaining module, and extract entities, entity attributes, and entity relationship knowledge therein; the knowledge fusion unit 202 is used for disambiguating and fusing multi-source knowledge to improve the knowledge quality; the knowledge processing unit 203 is used for deep mining based on the existing knowledge, and improving the breadth and depth of the knowledge; the knowledge updating unit 204 is used for updating and expanding the existing knowledge graph.
In one embodiment, the quality tracing algorithm module 300 includes an intelligent analysis unit 301, a quality tracing index unit 302, a tracing flow visualization unit 303, and an analysis report generation unit 304. The intelligent analysis unit 301 is configured to extract quality event related entity information from an input quality event to be queried, and map the quality event related entity information to a corresponding position of a quality tracing knowledge graph; the quality tracing index unit 302 is used for tracing quality problems, tracing from the node positioned by the intelligent analysis unit in the quality tracing knowledge graph along the entity relationship, comparing the node information recorded by the graph with the information of the input quality event, and finding out a responsible entity causing the quality event; the tracing process visualization unit 303 is used for visually displaying a quality problem tracing process and a quality problem propagation path; the analysis report generation unit 304 is configured to generate a quality tracing result analysis report according to the enterprise-defined template.
The nuclear power equipment quality tracing system provided by the invention can automatically extract knowledge from data resources related to nuclear power quality and perform fusion processing to form a knowledge graph for revealing the traceable association relation of quality characteristics, and update and expand the content of the knowledge graph through the knowledge. According to the input quality event, the related knowledge required by quality tracing stored in the knowledge map is utilized to drive the automatic tracing of the quality problem occurrence position along the quality forming path and the correlation analysis of the quality problem occurrence reason, so that the tracing of the quality event of the nuclear power equipment in the whole life cycle is realized. The system breaks through the traditional manual checking method in the nuclear power quality problem tracing process, comprehensively improves the quality tracing efficiency of nuclear power equipment, greatly reduces the cost and the personnel dependence, and realizes an effective and reliable guarantee technology of nuclear power 'safety quality loop'.
Yet another embodiment of the present invention provides a computer device comprising a processor and a memory for storing processor-executable instructions; wherein the processor is configured to: all steps in the nuclear power equipment quality tracing method based on the knowledge graph in the embodiment can be executed. Specifically, the processor, when executing the computer program, may implement the following steps:
in step S1, relevant data for quality tracing is acquired;
in step S2, a knowledge graph for quality tracing is constructed;
in step S3, a quality event trace back query is performed using a knowledge graph.
Yet another embodiment of the present invention provides a computer-readable storage medium having stored therein computer-executable instructions that, when executed by a processor, cause the processor to perform a method comprising:
in step S1, relevant data for quality tracing is acquired;
in step S2, a knowledge graph for quality tracing is constructed;
in step S3, a quality event trace back query is performed using a knowledge graph.
It will be understood by those skilled in the art that all or part of the processes of the methods of the above embodiments may be implemented by hardware related to computer program instructions, and the program may be stored in a computer readable storage medium, for example, in the storage medium of a computer system, and executed by at least one processor in the computer system, so as to implement the processes of the embodiments including the methods described above. Including but not limited to magnetic disk storage, optical disks, read-only memory, random access memory, and the like. In this case, the program code itself read from the storage medium can realize the functions of any of the above-described embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present invention.
The above embodiments are merely preferred embodiments of the present invention, which are not intended to limit the scope of the present invention, and various changes may be made in the above embodiments of the present invention. All simple and equivalent changes and modifications made according to the claims and the content of the specification of the present application fall within the scope of the claims of the present patent application. The invention has not been described in detail in order to avoid obscuring the invention.
Claims (10)
1. A nuclear power equipment quality tracing method is characterized by comprising the following steps:
s1: acquiring related data for quality tracing;
s2: constructing a knowledge graph for quality tracing;
s3: and performing quality event tracing inquiry by using the knowledge graph for quality tracing.
2. The quality tracing method for nuclear power equipment as claimed in claim 1, wherein the related data for quality tracing is derived from one or more of product quality defect data, product quality event related records, field standards and laws and regulations, quality plans, product supply relations, and product production and manufacturing process files.
3. The nuclear power equipment quality tracing method of claim 1, wherein step S2 further comprises the steps of:
s21: and (3) knowledge extraction: extracting the relationship among the entity nodes, the entity attributes and the entities in the related data for quality tracing to form a standardized knowledge representation form and storing the standardized knowledge representation form;
s22: and (3) knowledge fusion: performing fusion processing on the knowledge obtained by the knowledge extraction, eliminating ambiguity in the knowledge, eliminating wrong and redundant knowledge, and meanwhile, counting the frequency/probability of the entity appearing in all event spaces, and storing the frequency/probability as the attribute of a corresponding node;
s23: knowledge processing: reasoning according to the knowledge obtained by the knowledge fusion to generate new knowledge, wherein the new knowledge comprises the relationship and the entity attribute between the entities obtained through reasoning;
s24: and (3) knowledge updating: and (4) iteratively updating the knowledge spectrum content by steps S21-S23 by using related data which are newly added in the later period and are used for quality tracing.
4. The nuclear power equipment quality tracing method of claim 3, wherein the entity node in step S21 at least comprises: quality problem expression nodes, quality problem occurrence object nodes, quality problem occurrence reason nodes, nuclear power equipment and nuclear power equipment sub-component nodes, and nuclear power equipment/sub-component related phase nodes and phase responsibility entity nodes; the entity attributes at least include: quality standard and normal state attribute of the equipment/component and probability attribute of the entity node in the total event space; the relationships between entities include at least: the relation between equipment/components/related stages and responsible parties, the composition relation between equipment and equipment subcomponents, the incidence relation between quality problem expression and quality problem occurrence objects, the relation between quality problems and causes causing quality problems to occur and the circulation relation of equipment/component manufacturing and installing processes;
and/or the method for counting the occurrence frequency/probability of the entity in the whole event space in the step S22 comprises the following steps: by counting the quality problems recorded in different quality events, the probability/frequency of a certain entity appearing in all event spaces is obtained based on the statistical principle.
5. The nuclear power equipment quality tracing method of claim 1, wherein step S3 further comprises the steps of:
s31: intelligently analyzing the input quality event, and identifying the entity, entity relationship and quality problem expression information of the input quality event;
s32: mapping the information identified in S31 to nodes and relationships in the knowledge-graph;
s33: and driving search based on the nodes, attributes and relationship knowledge stored in the knowledge graph, and tracing the nodes and reason information generated by the input quality problem.
6. The nuclear power equipment quality tracing method according to claim 5, wherein the search in step S33 is: the method comprises the steps of positioning nodes, attributes and relations corresponding to quality problem state expressions in a quality tracing knowledge graph, reversely tracing objects with quality problems along a quality forming path of nuclear power equipment by using the nodes, attributes and relations stored in the knowledge graph, carrying out step-by-step comparison and association analysis according to the occurrence positions, quality problem state expressions and reasons of relevant historical quality problems and equipment normal state information, giving reasoning results of the reasons for the quality problems, calculating credibility according to probability weights of the nodes, judging the stages to which the quality problems belong, and forming a tracing chain of quality defect generation-phenomenon-object-related stages; the quality problem stage comprises the stages of design, purchase, construction and debugging of nuclear power equipment.
7. A nuclear power equipment quality tracing system is characterized by comprising:
the data resource acquisition module is used for acquiring data resource information required by quality tracing from various data sources;
the knowledge map construction module is used for processing various data acquired by the data resource acquisition module and extracting knowledge information in the data;
and the quality tracing algorithm module is used for searching the knowledge graph according to the input quality event drive, inquiring the quality problem generation node and possible reasons of the nuclear power equipment according to the position and quality state of the quality problem, and performing visual display.
8. The nuclear power equipment quality traceability system of claim 7, wherein the data resource acquisition module comprises:
the data acquisition unit is used for acquiring related data resources for quality tracing from a website, an enterprise KMS system and a platform database;
the data index unit is used for establishing an index for the data resources obtained by the data acquisition unit and classifying the data resources;
and/or the knowledge graph building module comprises:
the knowledge extraction unit is used for processing the data resources acquired by the data resource acquisition module and extracting the entity, entity attribute and entity relation knowledge in the data resources;
the knowledge fusion unit is used for carrying out disambiguation and fusion processing on the multi-source knowledge so as to improve the knowledge quality;
the knowledge processing unit is used for carrying out deep mining based on the existing knowledge and improving the breadth and depth of the knowledge;
the knowledge updating unit is used for updating and expanding the existing knowledge graph;
and/or the quality tracing algorithm module comprises:
the intelligent analysis unit is used for extracting quality event related entity information from the input quality events to be inquired and mapping the quality event related entity information to the corresponding position of the quality tracing knowledge graph;
the quality tracing indexing unit is used for tracing quality problems, tracing from the node positioned by the intelligent analysis unit in the quality tracing knowledge graph along the entity relationship, comparing the node information recorded by the graph with the information of the input quality event, and finding out a responsible entity causing the quality event;
the tracing process visualization unit is used for visually displaying the quality problem tracing process and the quality problem propagation path;
and the analysis report generating unit is used for generating a quality tracing result analysis report according to the established enterprise template.
9. A computer device, comprising: a processor and a memory for storing processor-executable instructions; the processor is configured to perform the method of any of claims 1-6.
10. A computer-readable storage medium having computer-executable instructions stored therein, which, when executed by a processor, cause the processor to perform the steps of the method of any one of claims 1-6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110816567.6A CN113487211A (en) | 2021-07-20 | 2021-07-20 | Nuclear power equipment quality tracing method and system, computer equipment and medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110816567.6A CN113487211A (en) | 2021-07-20 | 2021-07-20 | Nuclear power equipment quality tracing method and system, computer equipment and medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113487211A true CN113487211A (en) | 2021-10-08 |
Family
ID=77941572
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110816567.6A Pending CN113487211A (en) | 2021-07-20 | 2021-07-20 | Nuclear power equipment quality tracing method and system, computer equipment and medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113487211A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114324368A (en) * | 2022-03-10 | 2022-04-12 | 广东祥利科技有限公司 | Modified plastic production detection method and system based on mechanical vision detection |
CN114881474A (en) * | 2022-05-09 | 2022-08-09 | 山东大学 | Tire full life cycle quality tracing method and system based on knowledge graph |
CN115062918A (en) * | 2022-05-23 | 2022-09-16 | 冶金自动化研究设计院有限公司 | Slab quality tracing method based on rule engine and event reporting |
WO2023159574A1 (en) * | 2022-02-28 | 2023-08-31 | 西门子股份公司 | Anomaly detection method and apparatus, computer-readable medium and electronic apparatus |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112395424A (en) * | 2020-10-10 | 2021-02-23 | 北京仿真中心 | Complex product quality problem tracing method and system |
-
2021
- 2021-07-20 CN CN202110816567.6A patent/CN113487211A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112395424A (en) * | 2020-10-10 | 2021-02-23 | 北京仿真中心 | Complex product quality problem tracing method and system |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2023159574A1 (en) * | 2022-02-28 | 2023-08-31 | 西门子股份公司 | Anomaly detection method and apparatus, computer-readable medium and electronic apparatus |
CN114324368A (en) * | 2022-03-10 | 2022-04-12 | 广东祥利科技有限公司 | Modified plastic production detection method and system based on mechanical vision detection |
CN114324368B (en) * | 2022-03-10 | 2022-07-29 | 广东祥利科技有限公司 | Modified plastic production detection method and system based on mechanical vision detection |
CN114881474A (en) * | 2022-05-09 | 2022-08-09 | 山东大学 | Tire full life cycle quality tracing method and system based on knowledge graph |
CN115062918A (en) * | 2022-05-23 | 2022-09-16 | 冶金自动化研究设计院有限公司 | Slab quality tracing method based on rule engine and event reporting |
CN115062918B (en) * | 2022-05-23 | 2024-07-16 | 冶金自动化研究设计院有限公司 | Slab quality tracing method based on rule engine and event reporting |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN114579875B (en) | Equipment fault diagnosis and maintenance knowledge recommendation system based on knowledge graph | |
CN110889556B (en) | Enterprise operation risk characteristic data information extraction method and extraction system | |
CN110968699B (en) | Logic map construction and early warning method and device based on fact recommendation | |
CN112612902A (en) | Knowledge graph construction method and device for power grid main device | |
CN113487211A (en) | Nuclear power equipment quality tracing method and system, computer equipment and medium | |
CN111967761B (en) | Knowledge graph-based monitoring and early warning method and device and electronic equipment | |
US20200356726A1 (en) | Dependency graph based natural language processing | |
CN108959395B (en) | Multi-source heterogeneous big data oriented hierarchical reduction combined cleaning method | |
CN111427775B (en) | Method level defect positioning method based on Bert model | |
CN112395424A (en) | Complex product quality problem tracing method and system | |
CN109408804A (en) | The analysis of public opinion method, system, equipment and storage medium | |
CN113779272A (en) | Data processing method, device and equipment based on knowledge graph and storage medium | |
CN113987199B (en) | BIM intelligent image examination method, system and medium with standard automatic interpretation | |
CN113254507B (en) | Intelligent construction and inventory method for data asset directory | |
Kim et al. | Application of natural language processing (NLP) and text-mining of big-data to engineering-procurement-construction (EPC) bid and contract documents | |
Saini et al. | Automated traceability for domain modelling decisions empowered by artificial intelligence | |
Berko et al. | Knowledge-based Big Data cleanup method | |
Sun | A natural language interface for querying graph databases | |
Yin et al. | A deep natural language processing‐based method for ontology learning of project‐specific properties from building information models | |
CN117687824A (en) | Satellite fault diagnosis system based on quality problem knowledge graph | |
Nedumov et al. | Automation of data normalization for implementing master data management systems | |
Tang et al. | Constructing Aircraft Fault Knowledge Graph for Intelligent Aided Diagnosis | |
Razafindramintsa et al. | Logical services automatic location from eLEL | |
CN113505590A (en) | Method and system for identifying key data elements in database based on entity disambiguation | |
Belhadef | A new bidirectional method for ontologies matching |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |