Skip to main navigation Skip to search Skip to main content

A Two-Timescale Voltage Control Framework for Distribution Networks under Weather-Driven PV Variability

  • Chenxuan Gao
  • , Yidian Zhu
  • , Xian Zhang*
  • , Guibin Wang
  • *Corresponding author for this work
  • School of Robotics and Advanced Manufacture, Harbin Institute of Technology Shenzhen
  • Shenzhen University

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

With the growing penetration of distributed photovoltaics (PV), voltage regulation in active distribution networks becomes more challenging due to weather-driven variability of PV output. Rapid changes in solar irradiance can induce significant intra-hour fluctuations that are difficult to handle by conventional slow-response devices. To address this issue, this paper proposes a two-timescale weather-driven voltage control framework. In the slow timescale, weather evolution is explicitly modeled to construct PV output intervals, and a robust optimal power flow model is formulated to coordinate on-load tap changers (OLTCs) and capacitor banks (CBs). In the fast timescale, a multi-agent soft actor-critic (MASAC) algorithm is employed to provide real-time reactive power support from PV inverters, compensating for intra-hour PV fluctuations. Simulation results on a modified IEEE 33-bus system demonstrate that the proposed method effectively improves voltage profiles under various weather scenarios. In particular, under highly unstable conditions, the average voltage deviation is reduced by over 50%, highlighting the effectiveness of the proposed framework in handling fast PV variability.

Original languageEnglish
Title of host publication2026 IEEE 3rd International Conference on Electrical Power Systems and Intelligent Control, EPSIC 2026
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9798331552534
DOIs
StatePublished - 2026
Externally publishedYes
Event3rd IEEE International Conference on Electrical Power Systems and Intelligent Control, EPSIC 2026 - Hybrid, Tianjin, China
Duration: 22 May 202624 May 2026

Publication series

Name2026 IEEE 3rd International Conference on Electrical Power Systems and Intelligent Control, EPSIC 2026

Conference

Conference3rd IEEE International Conference on Electrical Power Systems and Intelligent Control, EPSIC 2026
Country/TerritoryChina
CityHybrid, Tianjin
Period22/05/2624/05/26

Keywords

  • multi-agent reinforcement learning
  • reactive power control
  • robust dispatch
  • voltage control
  • weather uncertainty

Fingerprint

Dive into the research topics of 'A Two-Timescale Voltage Control Framework for Distribution Networks under Weather-Driven PV Variability'. Together they form a unique fingerprint.

Cite this