CN-115221333-B - Knowledge graph construction method in numerical control programming field

CN115221333BCN 115221333 BCN115221333 BCN 115221333BCN-115221333-B

Abstract

The invention discloses a method for constructing a knowledge graph in the field of numerical control programming, which comprises the steps of numerical control programming knowledge collection and classification, numerical control programming field knowledge graph mode layer construction, numerical control programming field knowledge graph data layer construction, numerical control programming knowledge fusion and reasoning, numerical control programming knowledge graph knowledge storage and numerical control programming field knowledge graph visualization. The invention analyzes the representation form of knowledge in the numerical control programming field, provides a knowledge extraction method of various strategies, and effectively covers the whole numerical control programming field. Therefore, the knowledge graph in the numerical control programming field constructed based on the invention has larger model and more complete knowledge types. The invention utilizes the knowledge reasoning mode based on the ontology and semantic rules to discover the implicit semantic relation between the constructed numerical control programming knowledge graphs, ensures the knowledge quality of the constructed knowledge graphs and further expands the scale of the knowledge graphs in the numerical control programming field.

Inventors

LIU DAKUN
FANG XIFENG
GONG CHANYUAN
XIE CHENCHEN
ZHANG SHENGWEN
ZHANG CHUNYAN
SHEN JIE
ZHU PENGCHENG
ZHU CHENGSHUN

Assignees

江苏科技大学

Dates

Publication Date: 20260508
Application Date: 20220614

Claims (7)

1. The knowledge graph construction method in the numerical control programming field is characterized by comprising the following steps of: (1) The numerical control programming knowledge collection comprises process information collection, equipment state collection, personnel information collection, programming experience knowledge collection and programming case collection; (2) The method specifically comprises local ontology construction, CAM field general ontology construction and semantic rule construction; (3) The knowledge graph data layer construction in the numerical control programming field comprises a structured data knowledge extraction method based on relational database mapping, an unstructured text data knowledge extraction method based on NLP and a knowledge extraction method based on CAM numerical control programming cases; (4) The knowledge fusion and the knowledge reasoning of the local knowledge graph comprise the knowledge fusion of the local knowledge graph and the knowledge reasoning based on the general ontology and rules; (5) Knowledge graph knowledge storage in the field of numerical control programming; (6) Knowledge graph knowledge visualization in the field of numerical control programming; the step (2) specifically comprises the following steps: (2.1) constructing a local body, wherein the local body is a 4M1E body and comprises a Man local body, a Machine local body, a Material local body, a Method local body and an Environment local body, the Man local body is a local body built through numerical control programming personnel experience, operator capability, field expert advice and enterprise management personnel machining intention, the Machine local body is a local body built according to machining Machine tool information, numerical control system information, tool information and clamp information, the Material local body is a local body built according to machining characteristic information and Material information, the Method local body is a local body built according to a numerical control machining Method, a tool path Method, a cutting-in and cutting-out Method, a tool avoidance Method and a coordinate system setting Method, the Environment local body is a local body built according to workshop machining Environment information, and the Environment local body is particularly built according to a tool state, a Machine tool state and a working table working condition; (2.2) constructing a CAM numerical control programming field general body; (2.3) constructing numerical control programming semantic rules; in the step (3), the structured data knowledge extraction method based on the relational database mapping specifically includes: (3.1) respectively establishing a numerical control process information base, a workshop state information base and an operator information base by utilizing the collected numerical control programming knowledge; (3.2) establishing a relational database storage model, storing the data in a relational database, and setting the association relationship between the data by using the primary key and the external key of the relational database; (3.3) performing ontology relation mapping on the established database by utilizing the constructed local ontology and RDF database mapping protocol to obtain a programming dynamic information knowledge set in an RDF form; (3.4) constructing a numerical control programming text data set according to the unstructured numerical control programming knowledge obtained by collection; (3.5) preprocessing the text data, constructing a data set for triad extraction from the text data set, wherein after the text data set is preprocessed, each line in the text is taken as an extraction sample; Training an NLP entity recognition model and an NLP relation recognition model, and respectively training an NLP-based entity and relation algorithm extraction model by applying entity, relation classification results and a data set after text preprocessing; (3.7) performing triplet extraction on the text data set by applying the entity and relation recognition model obtained through training, wherein the result obtained through NLP entity extraction model extraction corresponds to the head node and the tail node of the triplet, and the result obtained through NLP relation extraction model extraction corresponds to the relation between the head node and the tail node of the triplet; (3.8) checking the extracted triplet knowledge by the local ontology in the application mode layer; And (3.9) constructing a numerical control programming experience knowledge graph by applying the verified triples.
2. The knowledge graph construction method in the digital control programming field according to claim 1, wherein in the step (3), the knowledge extraction method based on CAM digital control programming cases specifically comprises: (3.10) establishing a historical numerical control processing case library; (3.11) extracting machining feature information and numerical control programming operation in a historical numerical control machining case, wherein the machining feature information refers to geometric feature information, topology information and feature process information of machining features, and the numerical control programming operation refers to numerical control programming operation correspondingly established in CAM software according to the machining features; (3.12) carrying out cluster analysis on the historical processing characteristic information set, and finding out the most typical processing characteristic information unit in the enterprise historical processing characteristics by using the method; (3.13) extracting a center sample of each class as a typical processing characteristic information unit of the class characteristic according to the clustering result; (3.14) constructing a feature-operation information unit by utilizing the program operation in the CAM numerical control programming case with the feature information unit matching correspondence; (3.15) extracting knowledge of the feature-operation information unit by using the constructed local ontology; And (3.16) constructing a CAM numerical control programming case knowledge graph by using the extracted triplet relation.
3. The method for constructing a knowledge graph in the field of numerical control programming according to claim 1, wherein the step (4) is specifically: (4.1) local knowledge graph fusion, wherein the local knowledge graph fusion mainly refers to matching of entities in graphs, and whether the objects pointed by the entities in the same local knowledge graph or different local knowledge graphs are the same or not is judged through a probability model and a machine learning method so as to align the entities; And (4.2) applying the constructed universal ontology and the numerical control programming SWRL rule to carry out knowledge reasoning on the fused knowledge graph.
4. The method for constructing a knowledge graph in the field of numerical control programming according to claim 1, wherein the step (5) is specifically: (5.1) storing map knowledge extracted from dynamic programming information of the relational database in a temporary storage model; and (5.2) storing the numerical control programming knowledge graph obtained by the case extraction and the natural language text extraction in a persistent storage model.
5. The method for constructing a knowledge graph in the field of numerical control programming according to claim 1, wherein the step (6) is specifically: (6.1) constructing a query interface and inputting query content, developing a query interface in a front-end Web page, and inputting and receiving a user query request; analyzing query intention by utilizing natural language processing and regular expression in a back-end system developed by Java to analyze the query intention of a user and extract key information input by the user; (6.3) constructing a query statement and inquiring; (6.4) judging the query result, if the query result is not empty, sending the query result to the front end, and if the query result is empty, prompting that the query information does not exist; And (6.5) front-end visualization, namely sorting data received by the front end into a data format suitable for a visual plug-in unit so as to realize the visualization of the numerical control programming knowledge graph, and clicking a node in a front-end visual graph interface to generate a click event, wherein the event can automatically construct a query information statement of the node and generate an attribute information graph of the node.
6. A computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements a numerical control programming domain knowledge graph construction method as claimed in any one of claims 1 to 5.
7. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements a method of knowledge graph construction in the field of digitally controlled programming according to any one of claims 1-5 when executing the computer program.

Description

Knowledge graph construction method in numerical control programming field Technical Field The invention relates to a knowledge graph, in particular to a knowledge graph construction method in the field of numerical control programming. Background Although the integration of digital manufacturing systems such as CAD (Computer AIDED DESIGN)/CAM (Computer Aided Manufacturing) and the like enables the seamless connection of product design and manufacturing process, the collaborative parallelism of product development is realized, and the research and development period of products is effectively reduced. However, at present, some outstanding problems still exist in the field of numerical control programming, which are in serious limitation on the development period of products, and are specifically characterized in that the numerical control programming and preparing time of manufacturing enterprises is far longer than the processing time. This indicates that programming efficiency and quality have become key factors affecting product manufacturing cycle time. Factors limiting the numerical control programming efficiency and quality are mainly as follows: (1) The existing manufacturing systems of enterprises are poor in integration degree and semantic interoperability, the production flow documents of the enterprises are mostly stored in independent databases, and the multi-source heterogeneous data lack of standardized interfaces and unified representation methods, so that the management of the production flow documents is difficult and the process content is unclear. In addition, due to the existence of informationized island phenomenon between CAD/CAPP (Computer Aided Process Plan)/CAM, the management of design information, process information and manufacturing information is confusing, so that a digital programmer needs to spend a great deal of time to filter and extract the process information in the digital control programming stage. These factors cause problems such as low numerical control programming efficiency, and incapability of ensuring the quality of numerical control programs. (2) The numerical control programming system and the programmer are far away from the production field, so that the unnecessary reverse repair of the numerical control program is often caused by the fact that the actual production condition of a workshop is not known, the time is wasted, the efficiency is reduced, and even waste products are caused. In the conventional numerical control programming process, programmers need to select proper numerical control programming parameters for processing characteristics depending on personal cognition and production experience to design a reasonable feed route. For some complex part features, this approach to numerical control programming presents significant difficulties and challenges, mainly because numerical control programming is a knowledge-intensive task involving a series of decision-making activities, most of which are highly dependent on the long-term production experience knowledge accumulation of the programmer. (3) Enterprises accumulate a large number of CAM numerical control programming cases in the actual production process, and the cases are used as the most direct carriers of programming information, contain rich knowledge and experience and are valuable manufacturing resources of the enterprises. However, enterprises lack effective knowledge mining means, often rely on manual inquiry and retrieval to realize reuse of programming cases, resulting in low resource utilization rate and serious waste of resources. The knowledge graph is a semantic knowledge base of a directed graph structure, and is a semantic network for describing objective facts in a graph form. Where nodes of the graph represent entities or attributes and edges of the graph represent various semantic relationships between entities. The basic constituent units of the knowledge-graph are triplets in the form of "entity-relationship-entity". A group of triplets can describe a specific relationship, and the triplets connected with each other form a huge semantic network, so that complex relationships among things can be easily described. At present, knowledge graphs have been widely used in the internet field and have been primarily used in the industrial field by virtue of strong knowledge storage, knowledge retrieval and knowledge reasoning capabilities. Disclosure of Invention The invention aims to provide a knowledge graph construction method in the numerical control programming field, so that a numerical control programmer is assisted to carry out programming decision with a good visual effect, and further the complexity and the numerical control programming period of numerical control programming are effectively reduced. The technical scheme is that the knowledge graph construction method in the numerical control programming field comprises the following steps: (1) The numerical control programm