文档名：结合数据驱动与物理模型的主动配电网双时间尺度电压协调优化控制
摘要：高比例电动汽车、分布式风电、光伏接入配电网,导致电压频繁地剧烈波动.传统调压设备与逆变器动作速度差异巨大,如何协调是难点问题.该文结合数据驱动与物理建模方法,提出一种配电网双时间尺度电压协调优化控制策略.针对短时间尺度(min级)电压波动,以静止无功补偿器、分布式电源无功功率为决策变量,以电压二次方偏差最小为目标函数,针对平衡与不平衡配电网,基于支路潮流方程,计及物理约束构建了二次规划模型.针对长时间尺度(h级)电压波动,以电压调节器匝比、可投切电容电抗器挡位、储能系统充放电功率为动作,当前时段配电网节点功率为状态,节点电压二次方偏差为代价,构建了马尔可夫决策过程.为克服连续-离散动作空间维数灾,提出了一种基于松弛-预报-校正的深度确定性策略梯度强化学习求解算法.最后,采用IEEE33节点平衡与123节点不平衡配电网验证了所提出方法的有效性.
Abstract：Alargenumberofelectricvehicles(EVs),distributedsolarand/orwindturbinegenerators(WTGs)connectedtodistributionsystemsleadtofrequentandsharpvoltagesfluctuations.Theactionratesofconventionaladjustabledevicesandsmartinvertersareverydifferent.Inthiscontext,anoveldual-timescalevoltagecontrolschemeisproposedbyorganicallycombiningdata-drivenwithphysics-basedoptimization.Onfasttimescale,aquadraticprogramming(QP)forbalancedandunbalanceddistributionsystemsisdevelopedbasedonbranchflowequations.Theoptimalreactivepowerofrenewabledistributedgenerators(DGs)andstaticVARcompensators(SVCs)isconfiguredonseveralminutes.Whereas,onslowtimescale,adata-drivenMarkoviandecisionprocess(MDP)isdeveloped,inwhichthecharge/dischargepowerofenergystoragesystems(ESSs),statuses/ratiosofswitchablecapacitorsreactors(SCRs),andvoltageregulators(VRs)areconfiguredhourlytominimizelong-termdiscountedsquaredvoltagesmagnitudesdeviationsusinganadapteddeepdeterministicpolicygradient(DDPG)deepreinforcementlearning(DRL)algorithm.ThecapabilitiesoftheproposedmethodarevalidatedwithIEEE33-busbalancedand123-busunbalanceddistributionsystems.Thecontributionsofthispaperaresummarizedasfollows:(1)Combiningdata-drivenwithphysics-basedmethods,astrategyforcoordinatedcontroloffivedifferenttypesofadjustableequipment,namelyVRs,SCRs,ESSs,SVCsandDGsinvertersonfastandslowtimescalesisproposed.(2)Aslowtimescale(say1hour)MDPforactiveandreactivepowercoordinationisconstructed.The(near)optimalsettingsofratios/statusesofVRs,SCRsandcharge/dischargepowerofESSsarefoundusingDRLalgorithm.Asaresult,thedeficiencyoflowcomputingrateforconventionalphysics-basedlargescalemixedintegernon-convexnonlinearstochasticprogrammingiscompletelyovercome.(3)Thecharge/dischargepowerofESSiscontinuousvariablewhileratios/statuesofVRsandSCRsarediscretedecisions.DDPGalgorithmcannotbedirectlyapplicabletodiscreteactionwhileDQNalgorithmcannotbeapplicabletocontinuousaction.Further,whentherearealargenumberofVRsandSCRsindistributionnetwork,DQNalgorithmleadstodimensionalitycurses.TheexistingDRLalgorithminliteraturescannotdealwithjointcontinuous-discreteaction(efficiently).Toeliminatedimensionalitycursesinjointcontinuous-discreteactionspace,ratios/statusesofVRsandSCRsarefirstlyrelaxedtocontinuousvariables.Then,fortheprotoactiongivenbyactorofDDPGagent,Knnnearestneighborsarefoundoutinthejointcontinuous-discreteactionspace.Finally,eachoftheKnnactionsistransferredtothecriticofDDPGagentonebyonetoevaluateitsvalue.Theactionwiththegreatestvalueischosentointeractwiththedistributionnetwork.(4)Giventhe(near)optimalsolutionoftheslowtimescaleMDP,aQPforVVOwithDGs,SVCsinverterssettingsonfasttimescale(sayseveralminutesorseconds)tominimizesquaredvoltagesmagnitudesdeviationsaredevelopedforbalancedandunbalanceddistributionsystems.Asaresult,voltageviolationsonfasttimescalecausedbysizable,rapidandfrequentpowerfluctuationsfromrenewableDGsandfastchargedEVscanbemitigatedinrealtime.(5)Oneoftheoutstandingadvantagesoftheproposedmethodisveryeasytoperforminpracticewithnearoptimalsolution.Further,whenKnn=20orKnn=40,theproposedmethodhasmuchmorestabletrainingprocessthanmulti-agentDQNalgorithmandmuchhighercomputingratethanconventionalmulti-slotsinglefasttimescalemixedintegerQPbyabout18.0～36.7times.
作者：张剑 崔明建 何怡刚 Author：ZhangJian CuiMingjian HeYigang
作者单位：合肥工业大学电气与自动化工程学院合肥230009天津大学电气自动化与信息工程学院天津300072武汉大学电气与自动化学院武汉430072
刊名：电工技术学报
Journal：TransactionsofChinaElectrotechnicalSociety
年，卷(期)：2024, 39(5)
分类号：TM732
关键词：智能配电网 电压控制 深度强化学习 二次规划 双时间尺度
Keywords：Smartdistributionsystems voltagecontrol deepreinforcementlearning(DRL) quadraticprogramming(QP) dual-timescale
机标分类号：TM732TP391.9TM615
在线出版日期：2024年3月19日
基金项目：国家自然科学基金结合数据驱动与物理模型的主动配电网双时间尺度电压协调优化控制[
期刊论文] 电工技术学报--2024, 39(5)张剑 崔明建 何怡刚高比例电动汽车、分布式风电、光伏接入配电网,导致电压频繁地剧烈波动.传统调压设备与逆变器动作速度差异巨大,如何协调是难点问题.该文结合数据驱动与物理建模方法,提出一种配电网双时间尺度电压协调优化控制策略.针对短...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
结合数据驱动与物理模型的主动配电网双时间尺度电压协调优化控制 Dual Timescales Coordinated and Optimal Voltages Control in Distribution Systems using Data-Driven and Physical Optimization
结合数据驱动与物理模型的主动配电网双时间尺度电压协调优化控制.pdf
- 文件大小:
- 2.53 MB
- 下载次数:
- 60
-
高速下载
|
|
