Research On Water Pollution Event Tracing Method Based On Knowledge Graph Relational Reasoning

In recent years,due to factors such as land and water traffic accidents,accidental leakage and illegal discharges of factories and enterprises,man-made destruction,sudden extreme natural disasters and other factors,the frequent occurrence of water pollution incidents in China has caused people's lives and property safety and the balance of the water ecological environment system.A serious threat.Through water quality monitoring information of water bodies in watersheds,water pollution can be traced back quickly,effectively and accurately to obtain necessary historical information of pollution source items(pollution location,pollution volume and pollution transmission process),which can be used for risk assessment and emergency response of water pollution events Regulation,generation and disposal plans,and water environment supervision provide decision support,which will play a major role in the water pollution emergency regulation system.However,the existing methods and technologies for tracing water pollution incidents still face poor traceability and poor accuracy of traceability results,As well as various problems such as inadequate data utilization,the use of reliable and effective traceability methods for retrospective research on water pollution incidents is currently the main problem.In this paper,based on the analysis of traceability research on water pollution events through the coupled simulation model of river water quality at home and abroad,some deficiencies in the use of mechanism models for retrospective research are identified:(1)In the monitoring of water environment,the relative position and relative independence of different pollution sources Due to the limitations of scope of the industry's monitoring,it is difficult to expand beyond the monitoring scope,which makes the monitoring data have a certain lag;(2)Limited by the cognition of the statistical law of the mixing and diffusion of pollutants in river channels and water pollutants,the mechanism is still unclear and relevant Relationships describe difficult factors quantitatively and are difficult to consider in traceability methods.Based on this,this paper proposes a method for tracing water pollution events based on knowledge graph relational reasoning,and carries out research on the construction and application of knowledge maps for inland river water pollution tracing.The water pollution traceability knowledge map is a conceptual model that uses the domain knowledge map to integrate the knowledge of water pollution traceability for inland rivers.By constructing the water pollution traceability ontology,various knowledge related to water pollution based on the object dimension and the traceability process are involved formal rule expression of various temporal and spatial relationships and semantic relationships of retrospective reasoning;through the process of water pollution traceability knowledge map data layer construction method,knowledge extraction and fusion of various heterogeneous data collected by multiple sources are performed to extract expressions that meet the ontology rules Knowledge to form a unified water pollution traceability knowledge base;taking Panjiakou Reservoir and its upstream watershed as an empirical area,the abnormal water environment quality evaluation indicators in its water quality monitoring data are examples of analysis of water pollution events,and the body and knowledge of water pollution traceability Database,constructing water pollutant generation and transmission relationship chain and Panjiakou Reservoir and its upstream river basin instance knowledge map,using water quality section nodes as the starting and ending nodes for inference,referring to the transmission relationship of water pollutants,and completing water pollution events by matching nodes and paths The retrospective study and the follow-up monitoring report and remote sensing analysis results to verify the retrospective results of this article.The specific research content and results of the thesis include the following aspects:(1)Based on the research of water pollution incident tracing under the guidance of domain knowledge graph construction and application,first,from the definition of product production and logistics management perspective,the basic principles of water pollution tracing from the perspective of industry knowledge graph are proposed;second,It summarizes the classification and causes of water pollution events in inland rivers at home and abroad,and describes the research status of water pollution retrospective numerical simulation methods for different research topics.Finally,it introduces the development of domain knowledge maps in geographic information,security police,tourism,the construction and application status of other fields,and the research status of knowledge graph reasoning.Analyzes the deficiencies of traditional water pollution traceability methods based on water quality monitoring,hydrodynamics,pollutant transport and diffusion physical models and numerical simulation,and the advantages of industry knowledge map construction and reasoning in water pollution traceability scenarios.The main research content and overall ideas of this article are clarified.(2)Ontology construction of water pollution traceability knowledge map.Define and decompose knowledge in the field of water pollution tracing from the object dimension to form a single-dimensional and multi-layer knowledge system structure,which is used as a conceptual model of water pollution tracing model layer construction.Among them,water pollutants,sewage industry and pollution source concepts It is used to describe the generation and transmission relationship of water pollutants.Geographic information and hydrological information concepts are used to trace the activities and locations of water pollutants.In relational modeling,hierarchical and semantic relationships between concepts are modeled based on the water pollution traceability knowledge architecture,and geometric features and spatial relationships(directional relationship,topological relationship and measurement relationship)are carried out in geographic information and hydrological information concept classes based on geographic knowledge maps Relationship and measurement relationship)to form an ontological model of knowledge in the field of water pollution tracing and provide a regular description for the construction of the knowledge map of water pollution tracing.(3)Data layer construction of water pollution traceability knowledge map.In the field of water pollution traceability,entity and relationship knowledge element extraction,according to the characteristics of the data structure,for unstructured data,entity knowledge element extraction uses deep learning-based entity recognition method,and relationship knowledge element extraction uses deep learning-based relationship extraction method;For structured data,entity and relationship knowledge element extraction is converted and generated by D2 R knowledge mapping tools;for semi-structured data,in the data preprocessing,the Python web page analysis framework Beautiful Soup and regular expressions are used to extract entity and relationship knowledge elements.The extracted knowledge elements are used for knowledge fusion through ontology fusion tools.Finally,the fused knowledge is stored in the Neo4 j graph database in the form of graph structure to form a knowledge base of the water pollution traceability knowledge graph.(4)An example analysis of Panjiakou Reservoir and its upstream watershed based on the constructed water pollution traceability knowledge map.According to the constructed water pollution traceability knowledge map,the water pollutant generation and transfer relationship chain is proposed: "abnormal water environment quality evaluation indicators ? water pollutants ? sewage industry ? pollution source",and the method of tracing the source of pollution and the type of sewage industry is determined;Secondly,according to the connection relationship between the source and sink relationship graphs of the Panjiakou Reservoir and its upstream watershed,the watershed system map structure in the map is constructed,and the location of the pollution source is traced through node and path matching to complete the task of water pollution incident traceability research.The innovations of this article are as follows:(1)Based on the basic principles of water pollution tracing,the knowledge graph construction technology is used to effectively integrate the knowledge content and structure in the field of water pollution,and the water pollution tracing knowledge graph is constructed to carry out water pollution incidents in inland rivers.Retrospective study;(2)Inland river system(Panjiakou Reservoir and its upstream watershed)is divided into sub-basins,constructing watershed pollution source-sink relationship diagrams,and transforming it into a map structure through knowledge map construction technology,realizing the geographical structure of the river system Conversion of graph structure.The Panjiakou Reservoir and its upstream watershed are used as empirical areas for water pollution traceability,and the traceability results are verified through subsequent monitoring reports and remote sensing analysis results,affirming the application model and value of domain knowledge maps.