基于深度学习的农业无人机作物健康监测与数据挖掘研究

打开文本图片集

中图分类号：S127 文献标志码：A 文章编号：2096-9902（2026）04-0009-04

Abstract：Aimingattheproblemsofloweficiencyandinsuffcientaccuracyoftraditionalcrophealthmonitoring，an agriculturalUAVcrophealth monitoringsystembasedondeeplearning wasbuilt.Asix-rotorUAVequippedwithamult-spectral camera wasused toacquire wheatcanopyimages，andthelimprovedResNet-5O model wasusedtocarryout Automatic identificationandhealthclassficationofpestsanddiseases，andthespatio-temporaldistributionofcrohealthwereanalyzed throughassociationruleminingtechnology.TakingthejointingandflingstagesofwinterwheatintheNorthChinaPlainasthe testobject，thesystem'sacuracyinidentifying leaf diseasesreached92.3%，andthehealth indexgradingaccuracyeached 89.7% Theeficiencyofsingleinspectionwas15timeshigherthanthatofmanualinspection.Dataminingresultsrevealthatthe occurrenceofpowderymildewhasapositiecorrlationofO.76withsoilmoisture，andtheoccureneofstriperusthasanegative correlationofO.68withthedailyaveragetemperaturediferencetoprovidedatasupportforaccurateplantprotectiondecisions.

Keywords： deep learning; agricultural UAV; crop health monitoring; data mining; precision agriculture

作物健康监测是保障粮食安全的关键环节，传统人工巡检劳动强度大且时效性差，地面传感器空间覆盖受限，遥感影像分析依赖人工经验导致精度不稳定。（剩余5317字）