Research On Intelligent Knowledge Support For Urban Rail Transit Construction Safety Management

Urban rail transit construction project is a complex and high-risk systematic engineering,which has the characteristics of large-scale construction,large number of participants,complex technology,and changeable construction environment.It is easy to cause safety accidents.As safety accidents are the result of the interaction of various risk factors,safety management needs comprehensive knowledge support.Although the urban rail transit construction industry has accumulated a large amount of data,how to acquire the required knowledge quickly and accurately from a large number of data still lacks an effective solution when facing to the specific safety problems.In order to solve the above problems,this paper is based on URTCSM(Urban Rail Transit Construction Safety Management),and from the perspective of knowledge support.By introducing the relevant technology and methods in the field of artificial intelligence,it studies the theoretical model and method system of intelligent knowledge support for safety management based on knowledge graph.Specific contents include:Under the guidance of system theory,this paper analyses the core tasks and management processes of urban rail transit construction safety management.Then it puts forward the concept and connotation of intelligent knowledge support,and studies the knowledge graph technology in the field of artificial intelligence and its role in the knowledge support of urban rail transit construction safety management.On this basis,it constructs the theoretical model of intelligent knowledge support for urban rail transit construction safety management based on knowledge graph.This paper defines the domain knowledge scope of URTCSM,decomposes domain knowledge from process,organization,building and management,and forms a multi-dimensional hierarchical knowledge classification system.In domain conceptual modeling,based on the content and characteristics of domain knowledge architecture,a multi-dimensional hierarchical conceptual model of professional domain is constructed;a mixed-granularity conceptual model of standard specification is constructed according to its own structure and useage requirements;and a multi-agent related conceptual model of accident is constructed according to the needs of accident analysis for accident knowledge.In the aspect of entity relationship modeling,the hierarchical relationship between concepts is modeled based on domain knowledge classification architecture,and the relationship between the key elements affecting the safety implementation of urban rail transit construction project is modeled,forming the domain knowledge structure model of URTCSM,providing a standardized knowledge framework for the construction of domain knowledge graph.This paper analyses the main sources of knowledge in URTCSM,and focuses on the collection and organization of specifications and accident case data.In the field of domain entity knowledge element extraction,according to the degree of data structure and the characteristics of natural language description,different types of entity knowledge elements are extracted by manual extraction,transformation based on mapping relationship,rule-based extraction,and entity recognition based on deep learning.In the aspect of relational knowledge element extraction,mapping-based transformation,rule-based relational extraction,entity co-occurrence-based relational extraction and machine learning-based relational extraction are used respectively.Similar knowledge element extraction methods are used in entity's attribute recognition.The extracted knowledge elements need to be fused with the existing knowledge.By analyzing the needs of knowledge fusion in different situations,the corresponding fusion methods are proposed.At last,all kinds of entities and relational knowledge elements in knowledge graph are stored in Neo4 j graph database in the form of graph structure,and the URTCSM domain knowledge graph is established finally.A framework of intelligent knowledge support for URTCSM is proposed.Aiming at Specification knowledge,three methods of hybrid granularity specification knowledge acquisition are proposed: knowledge navigation,intelligent search and knowledge recommendation.Aiming at the application of accident knowledge,mainly supporting intelligent analysis of safety knowledge,three kinds of accident analysis tasks are proposed: displaying accident cognitive structure in the form of accident profiles,automatically constructing accident statistical analysis of query statements according to statistical analysis indicators,and depth analysis of accidents based on correlation paths.According to the relationship among knowledge elements in URTCSM domain knowledge graph,safety risks in different management situations are analyzed to provide knowledge support for safety risk identification and prevention.Finally,an intelligent knowledge support system based on URTCSM domain knowledge graph is developed,which can be used for maintenance and management of domain knowledge graph,intelligent acquisition of standard knowledge,intelligent analysis of safety incidents,and decision analysis of safety management.It provides an intelligent knowledge support platform for safety management decision-making.The paper has a total of 107 pictures,23 tables,and 209 references.