Expression And Inference Of Semantic Relationship In Fluvial Reservoir Architecture

For a long time,in the study of reservoir architecture,geologists have summarized various geological phenomena and mainly used natural language,geological map,pattern map and other methods to express the geometric shape,scale,direction and mutual relationship of the architecture.With the gradual development of geological research to the big data paradigm,the demand for automation,refinement,quantification and intelligence is increasing day by day.Aiming at the problems existing in the study of reservoir architecture language relationship at this stage,this paper takes fluvial reservoir architecture as the research object,conducts a systematic study around the expression,calculation and reasoning of architecture language relationship,and forms the completed theoretical and technical scheme.The main research results of this paper are as follows:(1)The semantic model of fluvial facies reservoir architecture is constructed.Based on the ontology method,the semantic model of fluvial facies reservoir architecture is constructed,the semantic entities commonly used in fluvial facies architecture research and the semantic relationship between entities are sorted out,and the fluvial facies reservoir architecture ontology is formed as the basic expression form of its semantic model.At the same time,the method of spatial scene analysis and instantiation is proposed through the semantic model,It lays a foundation for further scientific management of fluvial facies reservoir architecture information.(2)A mapping framework between semantic relations and basic spatial relations was established.To address the problem of automatic identification and inference of semantic relations,a method for computing basic spatial relations that support configuration semantic relations was proposed.The mapping between various semantic relations,such as topological,directional,and metric relations,and basic spatial relations was also presented.Based on the proposed mapping framework and relation computation method,quantitative calculation of semantic relations from the geometric features of configuration elements can be achieved,allowing for objective calculation of relations independent of subjective human knowledge.(3)A rule library for configuration semantic relation inference was constructed.To address the problem of rule construction for semantic relation inference,a reasoning approach based on qualitative spatial reasoning was proposed.Various semantic relation inverse reasoning and relation combination reasoning were implemented through rectangle algebra and combination table,leading to the formation of a rule library for semantic relation inference.(4)Typical semantic relations and distributed pattern reasoning.Based on the method of graph theory,the graph representation method of semantic relations and spatial distribution patterns is constructed,and the reasoning problem of semantic relations and distribution patterns is transformed into the mining problem of frequent relations and frequent subgraphs in the knowledge graph.In order to further analyze the spatial structure in the architecture,this paper proposes a graph model sampling method based on the deterministic relationship to obtain the state sequence set in the architecture to support the calculation of the transition probability matrix.Finally,by comparing the weighted degree in graph theory with the connectivity evaluation in profile,the evaluation methods of local and global connectivity are given from the perspective of graph model.(5)Case analysis of knowledge graph construction and pattern reasoning.In this paper,two profiles of Jurassic fluvial facies outcrops in Datong,Shanxi,and the wide valley section of the Yarlung Zangbo River from Qushui County to Naidong County in Tibet Autonomous Region are taken as experimental objects,respectively,to carry out knowledge graph construction and model reasoning experiments in the study area.According to the different concerns of modern sediment and outcrop profile architecture research and pattern mining,this paper designs the experimental process pertinently.It can be seen from the experimental results that on the one hand,this research has solved the problem of standardized expression of semantic relations in complex architecture scenes,and at the same time,it has realized objective typical semantic relations and pattern reasoning through the reasoning scheme based on knowledge graph.