江苏省县域城市性发展差异的BP神经网络测定.pdfVIP

2009年第4期 总第 108期 文章编号：1003—2398(2009)04—0038—05 江苏省县域城市性发展差异的BP神经网络测定 余 华 ’，张小林 1,2，黄飞飞 ’，崔开俊 ’，王丽莉 ’ (1．南京师范大学 地理科学学院，南京 210046；2．南京师范大学 区域发展与规划研究 中心，南京 210046) ANAI _sISoN URBANITYDEVELoPM暇NTDⅡFERENTIATIoNSoFCoI Ⅱ ARl S IN LANGSUPRoⅥ NCE BASED oN BPNEIJRAI．NETWoRK YUHua，ZHANGXiao—lin ，HUANGFei—fei，CUIKai-jun，WANGLi—li {1．CollegeofGeographicSciences,NanjingNormMUniversitX,Nanjing,210046,China；2．ResealCh CenterforRegionalDevelopmentPlanning,NanjingNormalUniversity,Nanjing210046,Cbna) Abstract： UrbanityreferstothemanifestationdegreeofurbannatureinacertainregiOn． whichshouldbe evaluatedcomprehensivelybyanintergatedindexsystem．Urbaniyt indexreferstothechoiceofdifferent comparison unitsand the determination ofdifferentevaluation criteria． ArtificialNeuralNetwork isa nonlinearsystem，andhasgoodfeaturesofself-organization， self-adaptionandself-leaming．Asoneofhte mostwidely-usednetworks， BPneuralnewt orktakesadvantagesoverfaulttolerance． robustnessand self-adaption．SothisPaDerbuildsnonlinearmodelofBPneuralnewtorktoavoidsubjectivityofindexweight tomeasureurbaniyt．Choosingfifty-wtocountiesinJiangsuprovinceassutdyobject，thispaperfirstselects 16represenattiveindicatorsfrom theaspectsofspaceconcentrationlevel， economicprorgesslevel， social developmentlevelandinrfastructuralfaciliyt constructionleve1． andconstructsindexsystem toevalua~ urbaniytcomprehensivevaluebyusingBPneuralnewtorktheoryandmethodaccordingtocomprehensive， represenattive，comparable，operationalandregionalprinciple，basedonstatisticdataofJiangsuprovincein 2005． Then， urbaniyt comprehensiveevaluationvalueoffifty-wtocounyt areasareclassifiedintofive degrees，includingthehighesturbaniyt，hihgerurbaniyt，middleurbaniyt，lowerurbaniytandthelowest urbnaiyt．Moreover，MapInf0softwareisappliedtoma