2025年面向电力系统暂态电压稳定性提升的深度反馈学习报告（英文版）.docxVIP

DeepFeedbackLearningforPowerSystemShort-TermVoltageStabilityEnhancement

LipengZhu(朱利鹏)

lpzhu@hnu.edu.cn

Contents

1

1ResearchBackground

2

2MainMethodology

3

3NumericalTest

4

4ConcludingRemark

2

1.ResearchBackground

□ClosingtheLooparoundDataAnalytics—CorrectiveControlTakingpromptandpropercountermeasurestostabilizethesystem

1.Data-DrivenBadPMU

2.Data-DrivenDynamicStability

3.Data-DrivenClosed-

PreprocessingCorrection

Assessment(DSA)

LoopCorrectiveControl

Semi-Supervised

RecurrentingBased

Spoality

Rrng

PriorKnowledgeBasedData-DrivenOnlineDSAOnlineCorrectiveControl

LabelingModelScheme

Real-TimePMUHistoricalInteractiveData

WAMS

DataPMUDataInterface

Closed-Loop

PracticalLarge-ScalePowerGrid

PMUData

Acquisition

Frameworkofdata-drivenpowersystemonlinedynamicsmonitoringandstabilitycontrol

3

4

1.ResearchBackground

□DRL-basedCorrectiveControlForSystemStabilityEnhancement

SimulaionResponsesPMUWAMSBasedMeasuremansSimul

SimulaionResponses

PMUWAMSBasedMeasuremans

SimulatadActicns

ActualActions

DeepReinforcementLearning

DataFlowinOffline

Learning

NumericalPowerGridSimulation

duringOnlineApplicntionRealityGap

PracticalLarge-ScalePowerGrid

(High-DimensionalState/ActionSpace)

SystemTranslentsObservation

AdaptiveControlScheme

(s,a,s?(reward)

DataFlow

a(action)

s(state)

DRL-basedcontrolschemeforlarge-scalepowersystem

St