DOI:10.11991/yykj.202305020网络出版地址:https://link.cnki.net/urlid/23.1191.u.20231220.1016.002基于深度学习的交通运输行业数据自动分级方法研究王继晔1,张少博2,叶润泽3,张绍阳21.陕西省交通运行监测中心,陕西西安7100752.长安大学信息工程学院,陕西西安7100643.西安电子科技大学党委组织部,陕西西安710071摘要:为促进交通运输行业信息系统互联、保障数据安全,进而推动行业健康发展,本文对交通运输行业中的数据共享和自动分级方法进行研究。本文基于卷积神经网络(convolutionalneuralnetworks,CNN)–双向门控循环单元(bidirectionalgatingrecurrentunit,BiGRU)–胶囊网络(capsulenetwork,CapsNet)模型的数据类别判定方法,设计并实现了交通运输信息资源目录系统中数据的自动分级模块,完成了行业领域下的数据自动分级。实验结果表明本文算法的准确率和F1值分别达到了70.48%和70.16%,明显高于现有的几种主流模型,可以有效提高数据分级的效率。关键词:交通运输行业;数据共享;数据安全;数据分级;深度学习;卷积神经网络;双向门控循环单元;胶囊网络中图分类号:TP399文献标志码:A文章编号:1009−671X(2024)02−0145−06ResearchonautomaticclassificationmethodoftransportationindustrydatabasedondeeplearningWANGJiye1,ZHANGShaobo2,YERunze3,ZHANGShaoyang21.ShaanxiProvincialTransportationOperationMonitoringCenter,Xi'an710075,China2.SchoolofInformationEngineering,Chang'anUniversity,Xi'an710064,China3.OrganizationDepartmentofTheCPCXidianUniversityCommittee,XidianUniversity,Xi'an710071,ChinaAbstract:Topromotetheinterconnectionofinformationsystemsinthetransportationindustry,ensuredatasecurity,andfurtherpromotehealthydevelopmentoftheindustry,thispaperstudiesdatasharingandautomaticclassificationmethodsinthetransportationindustry.Basedonthedataclassificationmethodofconvolutionalneuralnetwork-bidirectionalgatingrecurrentunit-capsulenetwork(CNN-BiGRU-CapsNet)model,thispaperdesignsandimplementstheautomaticdataclassificationmoduleinthetransportationinformationresourcedirectorysystem,andcompletestheautomaticclassificationofdataintheindustryfield.TheexperimentalresultsshowthattheaccuracyandF1scoreoftheproposedalgorithmhavereached70.48%and70...