Vol.1,No.2徐旻等院智能化无人机植保作业关键技术及研究进展
33
KeytechnologyanalysisandresearchprogressofUAV
intelligentplantprotection
1,21,21,21,21,2
MinXu,RuiruiZhang,LipingChen,QingTang,GangXu
渊1.100097,
2.100097
UAVplantprotectionoperationfacesverycomplicatedenvironmentalconditions.Ononehand,itsultralowaltitude
operationsarevulnerabletogroundstructuresandbasichydropowerfacilities;ontheotherhand,theeffectivenessofplant
protectionoperationisstrong,anditisnecessarytospraythepesticidestothespecificpartsofcropsattheprescribedtimesoasto
ensuregoodpesticideapplicationeffect.Atpresent,UAVplantprotectiontechnologymainlyreferstotheexistingmature
technologyandflightplatformingeneralaviationfieldtobasically"flyandspray".However,thelackofpenetratingresearchand
theoreticalguidanceonenvironmentalperceptioninfarmlandoperation,themovementmechanismofdropletsundertherotor
airflow,andthepenetrabilityofthedroplettodifferentcropscanopyleadtolowpenetrationrateoftheUAVplantprotection
operation,easydrifting,frequentaccidents,largedamageprobabilityandlowcomprehensiveoperationalefficiency.Benefiting
fromthebreakthroughsinartificialintelligence,parallelcomputingtechnologyandintelligenthardware,theUAVplantprotection
technologyisdevelopinginthedirectionofintellectualization,systematizationandprecision.Thereal-timeperceptionofthe
environmentundernonestablishedconditions,intelligentjobdecisionmethodbasedonintelligentrecognitionofcropdiseases
andpests,thecontrolofthetoward-targetpesticidesprayingcontrolbasedonthevariableofwindfielddropletdepositionmodel
andthedatabasedjobevaluationsystemhavegraduallybecomethekeytechnologyoftheUAVintelligentplantprotection.The
manuscriptanalyzedandsummarizedtheresearchstatusandtechnicalachievementsinthefieldofUAVintelligentplant
protectionfromthefieldinformationperception,themodelingandoptimizationcontrolofaccuratepesticideapplication,the
evaluationandmonitoringoftheoperationeffect.Basedontheexistingresearch,theresearchalsopredictedthedevelopment
trendofthekeytechnologiesofintelligentUAVplantprotectioninthefuture.Theclusteringmethodofhyper-spectralimage
acquisitionandcomputationalintelligencebaseddeeplearningrecognitionwillbecomethekeytechnologyforreal-timeand
efficientacquisitionofcroptargetinformationinplantprotectionwork,whichgreatlyimprovestheaccuracyofremotesensing
informationinversionrecognition;machinevisionandmultimachinecooperativesensingtechnologycanacquiredynamic
informationoffieldoperationatmultiplelevelsandtime;thehighprecisiondropletspectrumcontroltechnologyindependently
controlledbynozzledesignandtheprecisionvariablesprayingcontroltechnologybasedonthewindfieldmodelcanfurther
improvethedropletdepositioneffectandreducetheliquiddrifting;thebreakthroughofhighaccuracymeshsolutiontechnology
willchangethepredictionmodeofdropletdriftfromartificialexperiencejudgmenttocomputersimulationandnumerical
deduction;thejobpathplanningtechnologywillgreatlyimprovetheefficiencyofmultimachineandmultiareaoperationand
reducethedistanceofinvalidoperation;thejobqualityevaluationbasedonthereal-timedataofthesensorandtheoperation
supervisionsystemoflargedatatechnologywillreplacepeopletoeffectivelycontroltheprocessoftheUAVplantprotection
operation,achievedataandtransparencyofplantprotection,andensuretheprocessisobservableandcontrollable.
UAV;plantprotection;intelligence;sensing;spraying |
|
