Tomtit‐Raven Evolutionary Selector‐Reinforced Attention‐Driven: A High‐Performance and Computationally Efficient Cyber Threat Detection Framework for Smart Grids
Fuente:
Wiley Open Access
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Artículo Open Access |
| Veröffentlicht: |
Wiley
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1867015276160090112 |
|---|---|
| author | Priya R. Karpaga Praveen Kumar Balachandran Umawathy Techanamurthy Muhammad Ammirrul Atiqi Mohd Zainuri |
| author_facet | Priya R. Karpaga Praveen Kumar Balachandran Umawathy Techanamurthy Muhammad Ammirrul Atiqi Mohd Zainuri Priya R. Karpaga Praveen Kumar Balachandran Umawathy Techanamurthy Muhammad Ammirrul Atiqi Mohd Zainuri |
| collection | Wiley Open Access |
| contents | Tomtit‐Raven Evolutionary Selector‐Reinforced Attention‐Driven: A High‐Performance and Computationally Efficient Cyber Threat Detection Framework for Smart Grids Priya R. Karpaga Praveen Kumar Balachandran Umawathy Techanamurthy Muhammad Ammirrul Atiqi Mohd Zainuri Energy Science & Engineering ABSTRACT Identifying cyber threats maintains the security and operational stability of smart grid systems because they experience escalating attacks that endanger both operating data reliability and system stability and electricity grid performance. The current intrusion detection techniques struggle to deliver effective results because they have issues with improper feature selection along with many incorrect positive identifications and weak computational processing abilities. These issues prevent them from providing real‐time security in CPS. This research develops a new combined intrusion detection framework through Tomtit‐Raven Evolutionary Selector (TRES) for feature selection and Reinforced Attention‐Driven Transformer (ReAD‐Transformer) for intrusion classification tasks. The TRES model implements an evolutionary strategy based on natural principles to select features by optimizing discrimination along with reducing unwanted data in effective intrusion detection systems. ReAD‐Transformer employs innovative reinforced attention‐driven learning to recognize long dependencies as well as contextual patterns in network traffic data, which leads to precise anomaly detection. The developed approach utilizes deep learning‐based decision‐making to refine features while delivering high performance and accuracy through an efficient decision‐making process. Experimental testing reveals that the proposed system model delivers superior results with accuracy reaching 99.82% as well as precision of 99.10%, and recall score of 99.94% that exceeds state‐of‐the‐art current techniques with an F1‐Score of 99.50%. Testing and training aspects of this proposed method operate at the fastest rates of 38 and 0.3 s respectively which enables real‐time smart grid intrusion detection. The proposed TRES‐ReAD framework proves successful as a scalable high‐performance adaptive cybersecurity solution because it secures smart grid infrastructures from evolving cyber threats. 10.1002/ese3.70430 http://creativecommons.org/licenses/by/4.0/ |
| doi_str_mv | 10.1002/ese3.70430 |
| format | Artículo Open Access |
| id | wiley_oa_10_1002_ese3_70430 |
| institution | Wiley Open Access |
| license_str_mv | http://creativecommons.org/licenses/by/4.0/ |
| publishDate | 2025 |
| publisher | Wiley |
| record_format | wiley_oa |
| spellingShingle | Tomtit‐Raven Evolutionary Selector‐Reinforced Attention‐Driven: A High‐Performance and Computationally Efficient Cyber Threat Detection Framework for Smart Grids Priya R. Karpaga Praveen Kumar Balachandran Umawathy Techanamurthy Muhammad Ammirrul Atiqi Mohd Zainuri Energy Science & Engineering Tomtit‐Raven Evolutionary Selector‐Reinforced Attention‐Driven: A High‐Performance and Computationally Efficient Cyber Threat Detection Framework for Smart Grids Priya R. Karpaga Praveen Kumar Balachandran Umawathy Techanamurthy Muhammad Ammirrul Atiqi Mohd Zainuri Energy Science & Engineering ABSTRACT Identifying cyber threats maintains the security and operational stability of smart grid systems because they experience escalating attacks that endanger both operating data reliability and system stability and electricity grid performance. The current intrusion detection techniques struggle to deliver effective results because they have issues with improper feature selection along with many incorrect positive identifications and weak computational processing abilities. These issues prevent them from providing real‐time security in CPS. This research develops a new combined intrusion detection framework through Tomtit‐Raven Evolutionary Selector (TRES) for feature selection and Reinforced Attention‐Driven Transformer (ReAD‐Transformer) for intrusion classification tasks. The TRES model implements an evolutionary strategy based on natural principles to select features by optimizing discrimination along with reducing unwanted data in effective intrusion detection systems. ReAD‐Transformer employs innovative reinforced attention‐driven learning to recognize long dependencies as well as contextual patterns in network traffic data, which leads to precise anomaly detection. The developed approach utilizes deep learning‐based decision‐making to refine features while delivering high performance and accuracy through an efficient decision‐making process. Experimental testing reveals that the proposed system model delivers superior results with accuracy reaching 99.82% as well as precision of 99.10%, and recall score of 99.94% that exceeds state‐of‐the‐art current techniques with an F1‐Score of 99.50%. Testing and training aspects of this proposed method operate at the fastest rates of 38 and 0.3 s respectively which enables real‐time smart grid intrusion detection. The proposed TRES‐ReAD framework proves successful as a scalable high‐performance adaptive cybersecurity solution because it secures smart grid infrastructures from evolving cyber threats. 10.1002/ese3.70430 http://creativecommons.org/licenses/by/4.0/ |
| title | Tomtit‐Raven Evolutionary Selector‐Reinforced Attention‐Driven: A High‐Performance and Computationally Efficient Cyber Threat Detection Framework for Smart Grids |
| topic | Energy Science & Engineering |
| url | https://scijournals.onlinelibrary.wiley.com/doi/10.1002/ese3.70430 |