Research On The Construction And Application Of Knowledge Graph For Lifting And Hositing Operations Based On Multi-Source Data

The lifting and hoisting project is one of the most dangerous sub-projects.It has the characteristics of multi-person collaboration,strong professionalism,and easy to be affected by the surrounding environment.It is very prone to accidents,especially group deaths and injuries.Although a large amount of data has been accumulated in the field of lifting and hoisting operations,the messy and disorderly data and informationmakes it difficult for employees to quickly obtain comprehensive and accurate knowledge.Therefore,takeing lifting and hoisting operation as the starting point,this thesisuses natural language processing,deep learning and other related technologies and methods to study the construction of knowledge graph in the field of lifting and hoisting operation for multi-source data.And it is applied to the compliance review of special construction schemes,safety education and training,safety accident analysis and normative knowledge acquisition.The specific contents include :By analyzing the characteristics of knowledge structure in the field of lifting and hoisting operations and the needs of relevant employees for knowledge,the standard specifications and lifting accident investigation reports are determined as the main sources of knowledge.Then,based on knowledge engineering and knowledge graph technology,the method model and evaluation index of knowledge extraction in the field of lifting and hoisting operations are proposed.Through the analysis of lifting and hoisting operations process and multi-source data,the scope of knowledge in the field of lifting and hoisting operations are defined,and the professional knowledge in the field is decomposed from safety management,machinery,operation,organization and other dimensions to form a multi-dimensional hierarchical domain knowledge classification system.Based on the knowledge structure and content of domain knowledge classification system,the conceptual model of professional domain is constructed.Based on the characteristics and content of standard specification knowledge structure,the conceptual model of standard specification is constructed.Based on the structural characteristics and knowledge requirements of safety accident investigation report,the conceptual model of lifting accident is constructed.According to the domain knowledge classification system,the hierarchical relationship between concepts is constructed,and the non-hierarchical relationship between different concepts is modeled to form the knowledge structure model,which provides a theoretical basis for the construction of the knowledge graph of the listing and hoisting operations field.Collect and sort out standard specifications and accident investigation reports in the field of lifting and hoisting operations,and select different knowledge extraction methods according to the characteristics of multi-source heterogeneous data.For entity extraction,a system entity extraction method based on classification system,structured data,rules and deep learning is formed.For relation extraction,a relation extraction method based on text structure and rule is formed.For attribute extraction,similar entity extraction and relationship extraction methods are adopted.Based on the obtained entity-relationship-attribute triples,this thesis analyzes the methods of knowledge fusion in different stages and different situations,and proposes a knowledge storage strategy based on Neo4 j graph storage.Entities,relationships,and attributes are imported into the knowledge graph to form a knowledge graph in the field of lifting and hoisting operations,which provides an effective means for the integration and organization of domain safety management knowledge,and has the significance of promotion for other projects in construction field.Many aspects of application research are carried out on the knowledge map in the field of lifting operation.This thesis explores the framework of using domain knowledge graph to conduct compliance review of special construction schemes,studies the ways to realize safety education and training by using Neo4 j and knowledge base question and answer,and uses knowledge graph to complete safety accident analysis and normative knowledge acquisition,so as to provide multi-faceted knowledge support in the field of lifting and hoisting based on knowledge graph.There are 66 figures,26 tables and 119 references in this thesis.