Safe Reinforcement Learning for V2G-Enabled Electric Vehicle Aggregators

Ruben   Eland; S. Orfanoudakis; P.P. Vergara Barrios

doi:10.1007/978-3-032-19102-1_5

Safe Reinforcement Learning for V2G-Enabled Electric Vehicle Aggregators

Ruben Eland
, S. Orfanoudakis^*
, P.P. Vergara Barrios

^*Corresponding author for this work

Intelligent Electrical Power Grids

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

Abstract

The increasing penetration of Electric Vehicles (EVs) and renewable energy sources is placing significant stress on existing power grid infrastructure. This work investigates the application of vehicle-to-grid (V2G)-enabled smart charging in workplace environments from the perspective of EV aggregators, using real-world charging data from Dutch business parking lots. To address the limitations of conventional deep Reinforcement Learning (RL) methods in enforcing operational constraints, we propose a Safe RL method using the Constrained Variational Policy Optimization (CVPO) algorithm, specifically designed to reduce constraint violations and enhance reliability. Empirical results show that CVPO outperforms classic RL baselines and rule-based policies, closely approximating the performance of an optimal offline benchmark while exhibiting strong generalization to unseen scenarios.

Original language	English
Title of host publication	Machine Learning and Principles and Practice of Knowledge Discovery in Databases - International Workshops of ECML PKDD 2025, Revised Selected Papers
Editors	Irena Koprinska, João Mendes-Moreira, Paula Branco
Place of Publication	Cham
Publisher	Springer Nature
Pages	76-91
Number of pages	16
Edition	Part III
ISBN (Electronic)	978-3-032-19102-1
ISBN (Print)	978-3-032-19101-4
DOIs	https://doi.org/10.1007/978-3-032-19102-1_5
Publication status	Published - 2026
Event	European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2025 - Porto, Portugal Duration: 15 Sept 2025 → 19 Sept 2025

Publication series

Name	Communications in Computer and Information Science
Volume	2841
ISSN (Print)	1865-0929
ISSN (Electronic)	1865-0937

Conference

Conference	European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2025
Country/Territory	Portugal
City	Porto
Period	15/09/25 → 19/09/25

UN SDGs

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

SDG 7 Affordable and Clean Energy

Keywords

EV Aggregators
EV Smart Charging
Safe Reinforcement Learning

Access to Document

10.1007/978-3-032-19102-1_5

Cite this

Eland, R., Orfanoudakis, S., & Vergara Barrios, P. P. (2026). Safe Reinforcement Learning for V2G-Enabled Electric Vehicle Aggregators. In I. Koprinska, J. Mendes-Moreira, & P. Branco (Eds.), Machine Learning and Principles and Practice of Knowledge Discovery in Databases - International Workshops of ECML PKDD 2025, Revised Selected Papers (Part III ed., pp. 76-91). (Communications in Computer and Information Science; Vol. 2841 ). Springer Nature. https://doi.org/10.1007/978-3-032-19102-1_5

Eland, Ruben ; Orfanoudakis, S. ; Vergara Barrios, P.P. / Safe Reinforcement Learning for V2G-Enabled Electric Vehicle Aggregators. Machine Learning and Principles and Practice of Knowledge Discovery in Databases - International Workshops of ECML PKDD 2025, Revised Selected Papers. editor / Irena Koprinska ; João Mendes-Moreira ; Paula Branco. Part III. ed. Cham : Springer Nature, 2026. pp. 76-91 (Communications in Computer and Information Science).

@inproceedings{a58d9f40cddc4ddebb8be99ea1bfa8a8,

title = "Safe Reinforcement Learning for V2G-Enabled Electric Vehicle Aggregators",

abstract = "The increasing penetration of Electric Vehicles (EVs) and renewable energy sources is placing significant stress on existing power grid infrastructure. This work investigates the application of vehicle-to-grid (V2G)-enabled smart charging in workplace environments from the perspective of EV aggregators, using real-world charging data from Dutch business parking lots. To address the limitations of conventional deep Reinforcement Learning (RL) methods in enforcing operational constraints, we propose a Safe RL method using the Constrained Variational Policy Optimization (CVPO) algorithm, specifically designed to reduce constraint violations and enhance reliability. Empirical results show that CVPO outperforms classic RL baselines and rule-based policies, closely approximating the performance of an optimal offline benchmark while exhibiting strong generalization to unseen scenarios.",

keywords = "EV Aggregators, EV Smart Charging, Safe Reinforcement Learning",

author = "Ruben Eland and S. Orfanoudakis and \{Vergara Barrios\}, P.P.",

year = "2026",

doi = "10.1007/978-3-032-19102-1\_5",

language = "English",

isbn = "978-3-032-19101-4",

series = "Communications in Computer and Information Science",

publisher = "Springer Nature",

pages = "76--91",

editor = "Irena Koprinska and Jo{\~a}o Mendes-Moreira and Paula Branco",

booktitle = "Machine Learning and Principles and Practice of Knowledge Discovery in Databases - International Workshops of ECML PKDD 2025, Revised Selected Papers",

edition = "Part III",

note = "European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2025 ; Conference date: 15-09-2025 Through 19-09-2025",

}

Eland, R, Orfanoudakis, S & Vergara Barrios, PP 2026, Safe Reinforcement Learning for V2G-Enabled Electric Vehicle Aggregators. in I Koprinska, J Mendes-Moreira & P Branco (eds), Machine Learning and Principles and Practice of Knowledge Discovery in Databases - International Workshops of ECML PKDD 2025, Revised Selected Papers. Part III edn, Communications in Computer and Information Science, vol. 2841 , Springer Nature, Cham, pp. 76-91, European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2025, Porto, Portugal, 15/09/25. https://doi.org/10.1007/978-3-032-19102-1_5

Safe Reinforcement Learning for V2G-Enabled Electric Vehicle Aggregators. / Eland, Ruben ; Orfanoudakis, S.; Vergara Barrios, P.P.
Machine Learning and Principles and Practice of Knowledge Discovery in Databases - International Workshops of ECML PKDD 2025, Revised Selected Papers. ed. / Irena Koprinska; João Mendes-Moreira; Paula Branco. Part III. ed. Cham: Springer Nature, 2026. p. 76-91 (Communications in Computer and Information Science; Vol. 2841 ).

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Safe Reinforcement Learning for V2G-Enabled Electric Vehicle Aggregators

AU - Eland, Ruben

AU - Orfanoudakis, S.

AU - Vergara Barrios, P.P.

PY - 2026

Y1 - 2026

N2 - The increasing penetration of Electric Vehicles (EVs) and renewable energy sources is placing significant stress on existing power grid infrastructure. This work investigates the application of vehicle-to-grid (V2G)-enabled smart charging in workplace environments from the perspective of EV aggregators, using real-world charging data from Dutch business parking lots. To address the limitations of conventional deep Reinforcement Learning (RL) methods in enforcing operational constraints, we propose a Safe RL method using the Constrained Variational Policy Optimization (CVPO) algorithm, specifically designed to reduce constraint violations and enhance reliability. Empirical results show that CVPO outperforms classic RL baselines and rule-based policies, closely approximating the performance of an optimal offline benchmark while exhibiting strong generalization to unseen scenarios.

AB - The increasing penetration of Electric Vehicles (EVs) and renewable energy sources is placing significant stress on existing power grid infrastructure. This work investigates the application of vehicle-to-grid (V2G)-enabled smart charging in workplace environments from the perspective of EV aggregators, using real-world charging data from Dutch business parking lots. To address the limitations of conventional deep Reinforcement Learning (RL) methods in enforcing operational constraints, we propose a Safe RL method using the Constrained Variational Policy Optimization (CVPO) algorithm, specifically designed to reduce constraint violations and enhance reliability. Empirical results show that CVPO outperforms classic RL baselines and rule-based policies, closely approximating the performance of an optimal offline benchmark while exhibiting strong generalization to unseen scenarios.

KW - EV Aggregators

KW - EV Smart Charging

KW - Safe Reinforcement Learning

UR - https://www.scopus.com/pages/publications/105040254826

U2 - 10.1007/978-3-032-19102-1_5

DO - 10.1007/978-3-032-19102-1_5

M3 - Conference contribution

AN - SCOPUS:105040254826

SN - 978-3-032-19101-4

T3 - Communications in Computer and Information Science

SP - 76

EP - 91

BT - Machine Learning and Principles and Practice of Knowledge Discovery in Databases - International Workshops of ECML PKDD 2025, Revised Selected Papers

A2 - Koprinska, Irena

A2 - Mendes-Moreira, João

A2 - Branco, Paula

PB - Springer Nature

CY - Cham

T2 - European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2025

Y2 - 15 September 2025 through 19 September 2025

ER -

Eland R, Orfanoudakis S , Vergara Barrios PP. Safe Reinforcement Learning for V2G-Enabled Electric Vehicle Aggregators. In Koprinska I, Mendes-Moreira J, Branco P, editors, Machine Learning and Principles and Practice of Knowledge Discovery in Databases - International Workshops of ECML PKDD 2025, Revised Selected Papers. Part III ed. Cham: Springer Nature. 2026. p. 76-91. (Communications in Computer and Information Science). doi: 10.1007/978-3-032-19102-1_5