支持向量機在土壤水氮行為模擬中的應(yīng)用.pdf

1、分類弓榭級崎嚼蔗嘴犬等碩士學位論文單位代碼t10019學號：$040272支持向量機在土壤水氮行為模擬中的應(yīng)用ApplicationofSupportVectorMachinetoSimulationofSoilWaterandNitrogenBehavior研究生：拯紹整—————指導教卿：瑩霾值教援合作指導教師：申請學位門類級別：專業(yè)名稱：研究方向：所在學院：窶堂亟主土壤堂一一。j：鹱叢氯玨盞撼燃一童邀量登縫璺瞳一2006年6月Ab

2、stractInordertoacquirethesoilhydrauliccharacteristicparametersthatareneededinthesimulationofregionalsoilmoistureandsolutetransportPedoTmnsferFunctions(PTFs)modelsusingsupportvectormachineWereestablishedbasedonthesoilsurv

3、eydatainQuZhouCountyoftheNorthChinaPlainThemodelspredictedsaturatedhydraulicconductivityKs、saturatedwatercontent0s、residualwatercontent酢、thelogisticformofVanGenuchtanmodelparametersln(a1andInCh)Theinputvariablesincludeds

4、oiltexturebulkdensityandsoilorganicmattercontentThestatisticalresult“test)showedthatthepredictionvalueagreedwiththeobservationdatafairlywellandthesupportvectormachinemodelhadratherhighpredictionprecisionItwasalsofoundtha

5、twhiletheinputvariablesincludedsoiltexture，bulkdensityandsoilorganicmattercontentcallincreasedthepredictioncapabilitycomparingtothesimplifiedsoiltextureinputtreatmentOrsoiltextureandbulkdensityinputtreatment，orsoiltextur

6、ea嘲organicmatterinputtreatmentandthereiSnOobviousdifferencebetweenthepredictionprecisionofthesimplifiedsoiltextureinputtreatment，thesoiltextureandbulkdensityinputtreatment，orthesoiltextureandorganicmatterinputtreatmentBy

7、comparingthepredictingrelativeelTorthesupportvectormachineperformedbetterthanmultivariatestepwiseregressionmodel，butnotobviouslybetterthan轉(zhuǎn)PneurainetworkmethodMeanwhile，thesuppoavectormachinebasedOOtimeseriesWaSappliedto

8、predictthesoilmoisturecontentandnitrateni仃ogencontentThepredictionresultsshowedthatthepredictingerrorwasquitelargebutcomparingtothevariabilitywhichWaSmeasuredinfield，suchalargepredictingciterisacceptablethesuppo娃vectorma