Research On Location Optimal Distribution And Type Intelligent Selection Of Route Structures Based On Multi-dimensional Spatial Similarity Theory

The realization of intelligent design of railway route selection is an inevitable demand to adapt to the development of future route selection technology.Intelligent route selection design needs to comprehensively utilize modern information technology and simulate expert thinking to solve the technical problems encountered in the process of route selection.Modern route selection pays more attention to the layout and selection of major structures along the route,especially the selection of major bridge and tunnel locations and structural types often has a great impact on the overall plan decision of the route.How to use the experience accumulated in the long-term route selection process,based on the multi-dimensional spatial similarity theory,using computer simulation expert ideas,to achieve scientific decision-making of structure layout and selection has very important research significance.The corresponding research work is carried out from the aspects of designing case similarity criterion,constructing GIS case database,distributing railway structure location,selecting railway structure type and building railway 3D real scene model.The main research contents include the following aspects:(1)The case similarity criterion based on multi-dimensional spatial similarity theory and the construction method of railway structure case library based on GIS.Based on the theory of similarity science and similarity engineering,the multi-dimensional influencing factors such as economy,politics,technology,topography,geology,environmental sensitive area and road network planning affecting railway route selection are deeply analyzed.For different data types such as case numerical type and text type,the case similarity criterion based on multi-dimensional spatial similarity theory is designed.Based on the existing railway structure engineering information,professional engineer experience and technical specifications,combined with the designed case similarity criterion,the case is divided into attribute units and the storage rules are designed based on GIS to generate a self-learning,extensible,multi-source heterogeneous railway structure case database.(2)The optimal distribution algorithm of railway structure position based on multi-dimensional spatial similarity theory.Based on the idea of the whole life cycle of the railway,the unit price of the whole life cycle cost of the railway structure and the conversion coefficient of the unit price impact factor are calculated.Combined with the multi-dimensional spatial similarity theory,the method of estimating the unit price of the railway structure is designed.This method starts from the similarity of case attributes,by constructing SQL query expression in GIS,the GIS case database of railway structure is redeveloped,and the cost of similar cases is retrieved,reused and used to estimate the unit price of new cases.Taking the bridge erection height and tunnel excavation depth as independent variables,geometric constraints,structural constraints and geospatial environment constraints as constraints,and the minimum life cycle cost of railway structures as the objective function,a mathematical model for the optimal layout of railway structures is established,and an optimal location distribution algorithm of railway structures based on multi-dimensional spatial similarity theory is designed.(3)Integration of multi-dimensional spatial similarity theory and machine learning algorithms for intelligent selection of railway structure types.Based on the research of key technologies such as case representation,retrieval,reuse and intelligent design of railway structures,the type intelligent selection method of railway structures is designed by integrating multi-dimensional spatial similarity theory and machine learning algorithm.In order to achieve efficient retrieval of similar engineering cases,the nearest neighbor retrieval strategy is introduced into the similarity calculation of cases.Taking the retrieved existing cases as the sample data set,the BP neural network model is designed to simulate the human brain for training and learning,and the intelligent selection of railway bridge type is studied.At the same time,three regression prediction methods of decision tree,K-nearest neighbor and support vector machine are used for intelligent selection of railway bridges.Compared with the prediction accuracy of BP neural network algorithm,the accuracy and applicability of the prediction reasoning model are guaranteed,so as to realize the prediction reasoning and intelligent selection of railway structures.(4)Construction method of railway 3D real scene model based on ’ GIS + BIM ’Based on GIS technology,the comprehensive geographic information model of railway is established by integrating multi-source data such as route DEM data,DOM data,ground features,unfavorable geology and environmental sensitive area thematic information.Through Dynamo visual programming technology,the BIM model of subgrade,bridge,tunnel and ballasted track structure is established by secondary development of Revit.Based on the dynamic addition method,the overall matching algorithm between the three-dimensional environment model of railway structures and the BIM monomer model is designed.Based on the established three-dimensional real model of the railway,the practical applications of route design,route scheme review and optimization,conflict analysis and three-dimensional roaming are realized.