Research On Diversity And Robustness Of Urban Functional System Based On Multi-Source Data

With the development of society,there are various kinds of data related to human life,which often contain rich spatial and temporal information.Studying the relationship between multi-source data and urban functional system can provide data support for urban planning management departments.This thesis took Shenzhen as an example to study the diversity and robustness of urban functional system based on multi-source data.The main research contents and results of this paper are as follows.(1)Because communication base station data,point of interest data and traffic data have not been combined in urban function attribute inference,this thesis constructed a more extensive urban function system feature factor based on the above three kinds of data.The road network data was used to divide the study area,and the urban functional area attributes of the sample area were obtained by manual annotation.Through the research on the correlation between the characteristic factors of multi-source data and the attributes of urban functional areas,the best characteristic factors of urban functional areas were selected,and the multi-source feature data set of urban functional areas is obtained.(2)In view of the low accuracy of the research on the attribute inference of urban functional areas based on the supervised classification method,this thesis presented deep neural network(DNN)algorithm to carry out the research on the function inference of urban areas,and compared with the results of other machine learning algorithms to realize the monitoring and classification of urban functional areas based on deep neural network.The experimental results showed that the performance of the DNN model is the best,with the overall accuracy of 85.56 % and the kappa coefficient of 0.83.(3)In view of the lack of relevant research on the diversity analysis of urban functional system,this thesis,based on the attribute inference of urban functional areas by deep neural network,presented the diversity analysis indexes to carry out the diversity analysis of Shenzhen urban functional system.The results showed that the Shannon Wiener index and Simpson diversity index of Shenzhen urban functional system are 1.71 and 0.60,which are close to the ideal situation of the highest diversity of urban functional system.Therefore,the diversity of Shenzhen urban functional system is higher.(4)Because there is no relevant research on the robustness evaluation of urban functional system,this thesis presented the robustness evaluation method of traffic system network,designed the experimental simulation of Shenzhen City in different degrees and different ways of attack and destruction,and carried out the robustness evaluation of urban functional system according to the state change of Shenzhen city.The experimental results showed that the overall urban function system of Shenzhen city had poor robustness in face of targeted destruction,the change rate of fitting function was 1.25,and the proportion of network collapse points was 0.3.In the case of random destruction,the urban function system has strong robustness,the change rate of fitting function was 0.82,and the proportion of network collapse points was 0.6.Under the protection ratio of 10%,the robustness of urban functional system was improved better.For the overall urban function system,the change rate of fitting function decreased by about 28.01 %,and the proportion of network collapse points increased by about 54.59 %.