Research On Decision Model And Method Of Cartographic Generalization Of Residential Areas

Cartographic generalization is an inextricable problem in cartography.Cartography has the problem of generalization when it comes into being.Cartographic generalization is an important and core research topic in the field of cartography.It is a process of abstracting and simplifying the objective things and phenomena from the cognitive results of people to the production of maps.The process of abstracting and simplifying is bound to involve generalization and optimization.To realize the automation and intelligence of cartographic generalization is the unremitting goal in this field.It is not easy to make computer "understand" the rules of cartographic generalization which highly depend on abstract thinking,visual thinking and inspiration thinking.This is also the key reason why cartographic generalization is called "the world's problem" and "the most challenging problem".The academic circles at home and abroad have paid high attention to this problem for a long time.The process of cartographic generalization is essentially a decision-making process.The most common and critical information processing in cartographic generalization belongs to decision-making.In this paper,the problem of cartographic generalization is studied in the dimension of "decision".Mining knowledge from the principles of residential area selection,shape features and generalization,constructing decision-making model,design method,extracting features,quantifying features and carrying out experiments,to a certain extent,solve the problems of structural selection of residential area,pattern recognition of street network and shape maintenance of residential area.The main contributions and innovations of this paper are as follows:1.The key role of decision analysis in residential area synthesis is discussed.This paper focuses on the basic theory of residential land synthesis and decision analysis.This paper discusses that the key problems involved in residential land integration are decision-making problems,and puts forward the key role of decision-making analysis in residential land integration.This paper studies the development of decision theory and multi-attribute decision theory,the process of comprehensive decision-making of residential areas,and the main decision-making methods.The multi-attribute decision-making method based on information entropy,the principle of information entropy and the steps of calculating weight by entropy weight method are studied.By analyzing the calculation process of entropy weight method,the important properties of entropy weight method are obtained.2.The decision model and method based on knowledge are studied.The key of logical reasoning lies in rich domain knowledge.There are a lot of visual thinking,image thinking and inspiration thinking involved in residential area synthesis.It is difficult to solve these problems by logical reasoning or algorithm.The multi criteria decision-making model uses knowledge to comprehensively evaluate every natural state in the space,"select the best",which is in line with the principle of fuzzy processing in cartography It has obvious advantages to solve the problem with multi criteria decision model.This study provides a platform for the application of a large number of knowledge related to image thinking and inspiration thinking.Giving full play to people's thinking of image and inspiration,mining comprehensive knowledge of residential areas,designing comprehensive decision-making model,and using computer's efficient map information processing and map graphics processing ability are the specific research practice of human-computer integration mode.3.The key decision-making problems,models and methods in the structural selection of residential areas are studied.The key and difficult problem of residential area selection research lies in structured selection,and the essence of structured selection is decision-making.In the process of decision-making,it is necessary to comprehensively consider the hierarchical relationship,spatial relationship and topological relationship of residential areas,so as to determine which residential areas to select.In this paper,the general principles of residential land selection and cartographic specifications are fully considered.The decision-making method based on information entropy,combined with the system clustering method and Voronoi diagram spatial analysis method,is applied to the decision-making problems of point residential land and area residential land selection respectively,which can effectively overcome the artificial empowerment of existing methods and the limitations of existing explicit provisions of cartographic specifications It improves the scientificity of structured selection of residential areas.4.The key decision-making problems,models and methods of street network distribution pattern recognition are studied.The goal of residential land synthesis is to ensure the clarity of the map and to reflect the structural characteristics of the original residential land objectively.Street network is the skeleton of residential area,which determines the structure of residential area.Therefore,we must accurately judge the characteristics of the original street network.Street network can be classified into three shapes: grid,radial and irregular.Among them,the grid street network is one of the most widely distributed types with the most obvious characteristics.In this paper,based on a large number of measured map data,the characteristics of grid street network are analyzed,and decision models are designed for different scale street network.One is based on statistics,that is,the decision-making model of street network pattern recognition based on system clustering and coefficient of variation.The second is based on machine learning,that is,the decision-making model based on regional rectangle degree,straight line rate and rectangle rate.Neural network is used for decision-making.The experimental results of the two models are satisfactory.5.The key decision-making problems,models and methods of residential area shape generalization are studied.This paper focuses on the difficult problem of shape keeping in residential area generalization,designs the decision-making model and method,and carries on the experimental verification.There are three aspects: one is the decision-making model and method of street selection in grid residential area.The main idea is to propose long axis and short axis reference lines,quantitatively describe parallel streets,and try to keep parallel streets as much as possible.The second is the decision-making model and method of street selection in residential areas with irregular streets.The shape of this kind of residential area is mainly determined by the longer street line and larger Street mesh.This paper puts forward "perimeter attribute",that is to say,"perimeter attribute" is an important attribute of street selection decision-making,so as to achieve the purpose of correctly reflecting its contribution.The third is the decision-making model and method of building selection in non Street irregular residential area.The form of this kind of residential area is more complex,in fact,there is also a hidden skeleton line,that is,the distribution structure of the building.Taking the outline as the reference line,taking the distance from the outline as the clustering parameter,clustering the building,so as to distinguish different levels.In each level,the second clustering is carried out according to the distance,and the selection is carried out separately,so as to achieve the goal of retaining the skeleton and distribution density The purpose is to maintain the shape of residential area.