强化图注意力网络模型在选址路径问题中的应用

打开文本图片集

中图分类号：TP391 文献标志码：ADOI： 10.12475/aie.20250307

本文引用格式：，，.等.强化图注意力网络模型在选址路径问题中的应用[J].自动化与信息工程，2025，46（3）：43-51.HUANG Shuo， ZHANG Xuexi， XIE Xingwang， et al. Application of reinforced graph atention network model forlocation routing problem[J]. Automation & Information Engineering，2025，46（3）：43-51.

文章编号：1674-2605（2025）03-0007-09开放获取

Application of Reinforced Graph Attention Network Model for Location Routing Problem

HUANG Shuo ZHANG XuexiXIE XingWang ZHANG Tao （Guangdong University of Technology， Guangzhou 51ooo6， China）

Abstract：Toaddress the limitationof traditional atention network modelsinefectivelypreserving graph structural informationforLocationRoutingProblems，thispaperproposesareinforced graphatentionnetworkmodel.First，whileretaining graph structuralinformation，theencoderextracts node informationfromthe graph structurethroughatention mechanisms toobtain high-dimensional featurerepresentationsofnodesandglobal graph feature information.Then，thedecoderutilzes GatedRecurrent Units to efectivelycapture temporaldependencies innodesequencesandacquirescompletesolutions viastep-by-stepdecodng. Finaly，anauxiliaryValueNetwork isintroduced toevaluatethevalueofachaction，guiding policyupdatestoenhanctraiing efficiency.Experimentalresultsdemonstratethatthisreinforcedgraphatentionnetworkmodelcanrapidlyobtainhigh-quality solutions for LRP.

Keywords： location routing problem; reinforcement learing; graph atention network; value network; graph structure

0 引言

选址路径问题（location routing problem，LRP）是一个组合优化问题，旨在同时优化设施位置与车辆运输路径，以最小化总成本，广泛应用于物流、供应链管理等领域。（剩余10980字）