文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Prescriptive Treatment Optimization Using a Genetic Algorithm A Tool for Forest Management
侵权投诉

Prescriptive Treatment Optimization Using a Genetic Algorithm A Tool for Forest Management

Abstract: This paper describes research on the use of a multiobjective genetic algorithm (GA) to optimize prescriptive treatment plans for forest management. The algorithm is novel, in that (1) the plans generated by the algorithm are highly specific, stat

TheapproachdescribedinthispaperdiffersfromsimilarprojectsinpartbecauseaGAisusedtode-velopplanswithahighdegreeofspeci city.AframeworkhasbeendevelopedthatallowstheGAtoevolvehighlyspeci ctreatmentrecommendationsattheproject-levelforavarietyofmanagementgoals,timberandnon-timber(e.g.,visualorecologicalgoals)alike.Theapproachdoesmorethanlayoutahigh-levelstrategyforforestmanagement,ittellstheforesterwhen,where,andhowtoharvesteachstandinthemanagementunittomaximizethenum-berofmanagementgoalsthatcanbesatis edinthespeci edtimeperiod.

Theapproachisalsonovelinthattheplanspro-ducedbytheGAareevaluatedusinglogicbasedcomponentsoftheNED-2decisionsupportsystem(describedbelow).Eachplanisintendedtosatisfymultiplecompetinggoalswhicharespeci edusingknowledgebasesofNED-2,andtheirsatisfactionisdeterminedusingbuilt-ininferenceengines.The t-nessofeachindividualplanisbasedupontheresultsoftheseinferences.

Therestofthepaperdescribesthegeneticalgorithmanditsperformanceinmoredetail.Section1.2out-linesthespeci cation,simulation,andevaluationofplansinNED-2.Section2discussestherepresenta-tionoftreatmentplansasindividualsintheGAandintroducesthefunctionusedtoselectindividualsforsurvival.Section3presentstheexperimentsusedtoevaluatetheperformanceoftheGAandtheresultsobtained.1.1

TheNED-2DecisionSupportSystem

OurresearchontheGAbeganasanoffshootoftheNEDDecisionSupportSystemproject,acollab-orationbetweentheUniversityofGeorgiaandtheUSDAForestService.NED-2isadecisionsupportsystem(DSS)thatprovidesuserswithaplanningenvironmentforforestmanagementandexpertsys-temcomponentsforevaluatingandcomparingplanswithinamanagementunit(Nuteetal.[2004]).Inthiscontext,amanagementunitissimplytheareaofforestunderconsideration.Managementunitsaredividedintostands,whicharesectionsofforestwithsomeuniformcharacteristics.AplaninNED-2isascheduleoftreatments,suchasplantingorrowthinning,appliedtoindividualstandsoverpredeter-minedintervals.

TreatmentplansinNED-2aredividedintocycles,whichareseparatedbyregularintervals.EachplanissimulatedusingtheForestVegetationSimulator

(FVS),agrowthandyieldsimulatorproducedbytheUSDAForestService(doc.guandang.net).Atypicalplanmight,forexample,span ftyyearswithatreatmentperiodeverytenyears,foratotalof vecycles(managersmayalsoopttoselectnotreat-mentforacycleandjustlettheforestgrow).Theplaninthisexamplewouldconsistof vetreatments,chronologicallyordered.

TheNED-2knowledgebasecontainsrulesthatde- nedesiredfutureconditions(DFCs)forsatisfyingmanagementgoalsacrossdifferentforesttypes.Forexample,iftheuserselectsthegoal“Focusoncu-bicfootproduction”(atimbergoal),astandclassi- edasanaspen-birchforesttypemeetsthegoalifitstotalbasalareaandacceptablegrowingstockfortimberpasscertainthresholds.Thethresholdshavedifferentvaluesfordifferentforesttypes,depend-ingonhowproductiveonewouldexpectthemtobe.Rauscheretal.[2000]providesanextensivediscus-sionofthehierarchyofDFCsandgoalsinNED-2.Theserules,whilenotde nitive,haveasoundtheo-reticalbasisinexpertopinion.

TheDFCsareusedtoevaluatethesuccessorfail-ureofuser-de nedtreatmentplansinachievingasetofobjectives.Thepointistogivetheusersomefeedbackonhowlikelyhisorherplanistosucceedinmeetingtheirmanagementobjectivesaftertheyhavedesignedandsimulatedaplan.Goalsareas-signedafuzzy“con dencefactor”(CF)indicatingthedegreetowhichtheDFCrequirementsforagoalaresatis ed.ACFof1.0isassignedtogoalsthatarecompletelysatis ed,0.6tomarginallysatis edgoals,and0.4tonearlysatis edgoals.ACFof0.0indicatesthatthegoalcompletelyfailedtheassoci-atedDFCs.2DESIGNOFTHEGA2.1

Representation

Ageneticalgorithmingeneralattemptstosolveagivenproblembyevolvingpopulationsofcandidatesolutions.Newcandidatesarecreatedbycrossing(recombining)ormutatingmaterialfromexistingso-lutions.Inthiswaythespaceofpossiblesolutionsisexplored.Thecandidatesareevaluatedaccord-ingtosome tnessfunction,andrelatively tindi-vidualsaremoreoftenselectedtocreatenewindi-viduals.Overtime,thepopulationasawholecon-verges,hopefullytosomeoptimalsolution.Thechoiceof tnessfunctionandmutation,recombina-tion(crossover),andselectionoperatorshasadra-maticeffectontheperformanceofthealgorithm.

上一页第2页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email