Two-Stage Deep Reinforcement Learning for Distribution System Voltage Regulation and Peak Load Management: Preprint

Yansong Pei; Yiyun Yao; Junbo Zhao; Fei Ding; Jiyu Wang

Two-Stage Deep Reinforcement Learning for Distribution System Voltage Regulation and Peak Load Management: Preprint

Yansong Pei, Yiyun Yao, Junbo Zhao, Fei Ding, Jiyu Wang

Power Systems Engineering

University of Connecticut

Research output: Contribution to conference › Paper

Abstract

The growing integration of distributed solar photovoltaic (PV) in distribution systems could result in adverse effects during grid operation. This paper develops a soft actor critic-based deep reinforcement learning (SAC-DRL) solution to simultaneously control PV inverters and battery energy storage systems for voltage regulation and peak load demand shaving. The novel two-stage framework, featured with two different control agents, is applied for daytime and nighttime operation to enhance the control performance. Comparison results with other control methods on a real feeder in Western Colorado demonstrate that the proposed method can provide advanced voltage regulation with modest active power curtailment for peak demand reduction.

Original language	American English
Number of pages	8
State	Published - 2023
Event	2023 IEEE Power & Energy Society General Meeting - Orlando, Florida Duration: 16 Jul 2023 → 20 Jul 2023

Conference

Conference	2023 IEEE Power & Energy Society General Meeting
City	Orlando, Florida
Period	16/07/23 → 20/07/23

NREL Publication Number

NREL/CP-5D00-84637

Keywords

deep reinforcement learning
distribution system
peak load management
voltage regulation

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

https://www.nrel.gov/docs/fy23osti/84637.pdf

Cite this

@conference{94d7db861a114bb5b5b3807557289fda,

title = "Two-Stage Deep Reinforcement Learning for Distribution System Voltage Regulation and Peak Load Management: Preprint",

abstract = "The growing integration of distributed solar photovoltaic (PV) in distribution systems could result in adverse effects during grid operation. This paper develops a soft actor critic-based deep reinforcement learning (SAC-DRL) solution to simultaneously control PV inverters and battery energy storage systems for voltage regulation and peak load demand shaving. The novel two-stage framework, featured with two different control agents, is applied for daytime and nighttime operation to enhance the control performance. Comparison results with other control methods on a real feeder in Western Colorado demonstrate that the proposed method can provide advanced voltage regulation with modest active power curtailment for peak demand reduction.",

keywords = "deep reinforcement learning, distribution system, peak load management, voltage regulation",

author = "Yansong Pei and Yiyun Yao and Junbo Zhao and Fei Ding and Jiyu Wang",

year = "2023",

language = "American English",

note = "2023 IEEE Power & Energy Society General Meeting ; Conference date: 16-07-2023 Through 20-07-2023",

}

TY - CONF

T1 - Two-Stage Deep Reinforcement Learning for Distribution System Voltage Regulation and Peak Load Management: Preprint

AU - Pei, Yansong

AU - Yao, Yiyun

AU - Zhao, Junbo

AU - Ding, Fei

AU - Wang, Jiyu

PY - 2023

Y1 - 2023

N2 - The growing integration of distributed solar photovoltaic (PV) in distribution systems could result in adverse effects during grid operation. This paper develops a soft actor critic-based deep reinforcement learning (SAC-DRL) solution to simultaneously control PV inverters and battery energy storage systems for voltage regulation and peak load demand shaving. The novel two-stage framework, featured with two different control agents, is applied for daytime and nighttime operation to enhance the control performance. Comparison results with other control methods on a real feeder in Western Colorado demonstrate that the proposed method can provide advanced voltage regulation with modest active power curtailment for peak demand reduction.

AB - The growing integration of distributed solar photovoltaic (PV) in distribution systems could result in adverse effects during grid operation. This paper develops a soft actor critic-based deep reinforcement learning (SAC-DRL) solution to simultaneously control PV inverters and battery energy storage systems for voltage regulation and peak load demand shaving. The novel two-stage framework, featured with two different control agents, is applied for daytime and nighttime operation to enhance the control performance. Comparison results with other control methods on a real feeder in Western Colorado demonstrate that the proposed method can provide advanced voltage regulation with modest active power curtailment for peak demand reduction.

KW - deep reinforcement learning

KW - distribution system

KW - peak load management

KW - voltage regulation

M3 - Paper

T2 - 2023 IEEE Power & Energy Society General Meeting

Y2 - 16 July 2023 through 20 July 2023

ER -

Two-Stage Deep Reinforcement Learning for Distribution System Voltage Regulation and Peak Load Management: Preprint

Abstract

Conference

NREL Publication Number

Keywords

UN SDGs

Access to Document

Fingerprint

Cite this