文档摘要：电-气互联系统调度决策问题旨在实现天然气系统和电力系统中可调节资源的最佳配置,其精准性与高效性直接影响电-气互联系统运行的安全性与经济性.为描述可调节资源离散状态、非线性运行特性等物理性质,电-气互联系统调度决策问题中含有规模庞大的离散决策变量,模型复杂度高,使得现有依赖于商业混合整数线性规划(MILP)求解器的电力系统运筹优化技术面临"组合爆炸"的计算负担.为此,该文提出一种分支定界搜索信息深度引导的电-气互联系统调度决策加速求解方法.所提方法利用分支定界初始搜索阶段的信息构建小规模辅助MILP模型,并内嵌于分支定界搜索过程,引导剪除更多冗余搜索空间,在不损失最优性的前提下加速收敛.基于RTS-GMLC电力系统和天然气系统不同负荷水平及线性分段数下的30个算例仿真结果说明,相比于直接使用商业MILP求解器,所提方法在不损失最优性的前提下可实现平均4.20倍的加速,验证了所提方法的有效性.
Abstract：Thedispatchprobleminintegratedelectricityandgassystemisformulatedasthemixed-integerlinearprogramming(MILP)formtodescribethediscretefeatureofresources(suchastheunitstatuses)andnonlinearoperationrules(suchasthepowerflowequation,theWeymouthequation,etc.).Theoptimalsolutionachievesthebestallocationofresourcesandindicatesthesecurityandeconomicoftheintegratedelectricityandgassystem.However,thelarge-scaleintegervariablesbringthe"combinatorialexplosion"challenge,evenforthestate-of-the-artcommercialsolvers.Toaddressthisproblem,existingresearchfocusesontheexternalalgorithms,suchasreformulating,reducingthescaleofconstraints/integervariables,etc.However,itishardtobalancethesolutionefficiencyandtheerrorboundguaranteesinpractice.Therefore,thispaperproposesaninternalalgorithmthatishighlycombinedwiththesearchinformationinthebranch-and-boundprocess.Thedistinctadvantageoftheproposedmethodisthattheuniquestructureofintegratedelectricityandgassystemisconsideredalongwiththeabundantinformationduringthesolutionprocess.Asaresult,theproposedmethodcanachieveanaccelerationwithoptimalityguaranteed.ThispaperreviewstheMILPformulationofthedispatchprobleminintegratedelectricityandgassystem,andfocusesonthecomputationalbottleneck,i.e.,thelarge-scaleintegervariablesintroducedbythepiecewiselinearizationstructureoftheWeymouthequation.Becausemostofintegervariablesinthepiecewiselinearizationstructureremainszero,thispaperproposestoevaluatethepotentialeffectiverangeofintegervariablesintheoptimalsolution,basedonthebranch-and-boundsearchinformation.First,theproposedmethodcollectstherelaxationsolutionsduringtheinitialstageofthebranch-and-boundprocess,tobuildasearchinformationdatasetthatimpliestheoptimalvalueofthepiecewiselinearizationstructure.Therelaxationsolutionsareeasytoobtain,andtheyprovidelowerboundstothecurrentsearchtree.Therefore,theideaisthatifanintegervariableofthepiecewiselinearizationalwaysliesinasimilarrangeacrosstheabundantrelaxationsolutions,itislikelytoremainthesamepatternintheoptimalsolution.Second,thek-nearestneighborsregressionalgorithmisusedtoestimatethepotentialeffectiverangeofthepiecewiselinearizationstructure.Third,theproposedmethodbuildsasmall-scaleMILPmodelandsolvesitinparalleltoproduceabettersolutionthanthemainbranch-and-boundprocess.Asaresult,theredundantsearchspaceinthesearchtreeisprunedbythebettersolutionwithnoaccuracylossandtheoptimalityisguaranteed.Inthecasestudy,theeffectivenessoftheproposedmethodhasbeendemonstratedinseveralaspectsbasedontestcasesofRTS-GMLCandgassystem.First,fordifferentscalesofpiecewiselinearizationstructures,comparedwiththecommercialsolver,theproposedmethodacceleratesthedispatchproblemfrom1.60timesto36.90times(11.28timesonaverage).Theresultalsoshowsthatwiththeimprovementofthepiecewiselinearizationscale,thecomputationalburdensuddenlyincreasesandcanbesignificantlymitigatedbytheproposedmethod,whichprovidesabalancebetweentheformulationprecisionandthecomputationalefficiency.Second,for30testcasesusingdifferentpowerdemands,theproposedmethodacceleratesthedispatchproblemfrom1.64timesto13.52times(4.20timesonaverage).Itshowsthattheproposedmethodbehaveswellincomplicatedoperationconditions.Third,thecasestudydiscussessomedetailsoftheproposedmethod,includingthefeasibilityrepair,thepredictionaccuracy,theupperboundconvergence,thesearchtreescale,andtimecostofeachstep,whichprovidesathoroughlyanalysisfortheproposedmethod.Finally,atestcaseisusedtodemonstratetheeffectivenessinthelarge-scalesystem.Inconclusion,thispaperproposesanaccelerationalgorithmforthedispatchprobleminintegratedelectricityandgassystem.Theproposedmethodimprovesthesolutionefficiencywithoutthelossoftheoptimalityguarantees,whichispromisinginpracticalandprovidesanovelviewontheoptimizationinpowersystems.
作者：高倩 杨知方 李文沅 卢毓东 Author：GaoQian YangZhifang LiWenyuan LuYudong
作者单位：输变电装备技术全国重点实验室(重庆大学)重庆400044国网浙江省电力有限公司电力科学研究院杭州310014
刊名：电工技术学报 ISTICEIPKU
Journal：TransactionsofChinaElectrotechnicalSociety
年，卷(期)：2024, 39(13)
分类号：TM73
关键词：电-气互联 调度决策 混合整数线性规划 加速算法
Keywords：Integratedelectricityandgassystem dispatch mixed-integerlinearprogramming accelerationalgorithm
机标分类号：TM711TP391.41TU427.2
在线出版日期：2024年7月22日
基金项目：国家电网有限公司总部科技项目分支定界搜索信息深度引导的电-气互联系统调度决策加速求解方法[
期刊论文] 电工技术学报--2024, 39(13)高倩 杨知方 李文沅 卢毓东电-气互联系统调度决策问题旨在实现天然气系统和电力系统中可调节资源的最佳配置,其精准性与高效性直接影响电-气互联系统运行的安全性与经济性.为描述可调节资源离散状态、非线性运行特性等物理性质,电-气互联系统调度...参考文献和引证文献
参考文献
引证文献
本文读者也读过
相似文献
相关博文
关键词：电-气互联,调度决策,混合整数线性规划,加速算法,
- 文件大小:
- 16.33 MB
- 下载次数:
- 60
-
高速下载
|
|
1
