An evolutionary deep reinforcement learning-based framework for efficient anomaly detection in smart power distribution grids

Mohammad Mehdi Sharifi Nevisi; Mehrdad Shoeibi; F. Hernando-Galego; Diego Martín; Sarvenaz Sadat Khatami

doi:10.3390/en18102435

Publication record · 2025

An evolutionary deep reinforcement learning-based framework for efficient anomaly detection in smart power distribution grids

Authors: Mohammad Mehdi Sharifi Nevisi · Mehrdad Shoeibi · F. Hernando-Galego · Diego Martín · Sarvenaz Sadat Khatami
Year: 2025
Publisher: MDPI
Journal or publication: Energies
Google Scholar citations: 24

Journal indicators

CiteScore 2025 (Scopus)8.3

Best quartileQ1

Top 10% in any categoryYes

Positions by category

Control and Optimization10/198 · Q1 · percentile 95 · Top 10%
Engineering (miscellaneous)41/300 · Q1 · percentile 86
Electrical and Electronic Engineering160/1030 · Q1 · percentile 84
Energy Engineering and Power Technology68/347 · Q1 · percentile 80
Fuel Technology37/138 · Q2 · percentile 73
Renewable Energy, Sustainability and the Environment107/345 · Q2 · percentile 69
Energy (miscellaneous)35/106 · Q2 · percentile 67

Scopus source ↗

JCR (Clarivate): Not verified: the supplied ranking file is Scopus CiteScore, not JCR.

Included in the supplied report on non-standard bibliometric behaviour (2017–2019).

The report is an independent analysis published in 2021; it is not an official ANECA ban or endorsement list.

Research summary

A deep-reinforcement-learning anomaly detector combines CNN feature extraction, recurrent temporal modelling and artificial-bee-colony hyperparameter optimization. Four smart-power datasets are used to evaluate generalization, false alarms and computational efficiency.

Original description prepared for this website; consult the publication for its authoritative abstract.

DOI 10.3390/en18102435 ↗ · Open-access version ↗ · View Google Scholar record ↗