Study On The BP Neural Network And SVM Model For Establishing Xi'an Gaoling Urban Growth Boundary

By delineating the urban growth boundary(UGB),foreign countries controlled the urban development,improved the level of the economical and intensive use of urban land and protected the cultivated land successfully.Inspired by the successful examples,China began to set up the national UGB pilot work in 2014,hoped that to optimize the urban space layout,prevent the disorderly overspread of the city and enhance the efficiency of construction land use by setting UGB.This paper takes Gaoling District of Xi'an as the research area,setting the UGB with the space delineate on the basis of GIS and RS.Firstly,based on the remote sensing images of Gaoling area in 2010 and in 2015,the Normalized Differences Berrn Index(NDBI)was used to extract the boundary of urban(town),and the 360 points were extracted to approximate the city(town)boundary;Secondly,According to the actual circumstance of Gaoling,select 8 factors based on the “the six line of city” rigid control concept,such as elevation,roads,basic farmland,etc.,to obtain the value of influence factors;Finally,Based on the R software,the BP neural network and Support Vector Machine(SVM)technology in spatial data mining are used to establish the model of scaling factor influencing factors and distance,which can be used to delineate UGB in Gaoling District.Comparing the demarcated results of these two kinds of model,because of the differences of the structure of the models,the prediction results of different directions have their own advantages and disadvantages.Therefore,the same direction error correction method is used to model the UGB.The analysis of the combined model shows that the expansion direction of the delineated city(town)in line with the economic development in Gaoling,which can provide a reference for revising and perfecting urban and general land use planning.