Scholarlink

<p>The rapid integration of electric vehicle charging stations (EVCSs) into the distribution networks introduces the highly unbalanced load patterns which creates the fertile ground for cyber-attacks like load redistribution (LR) attacks. In order to identify load redistribution (LR) attacks, this paper proposes a hybrid approach that combines deep learning and mathematical modelling to ascertain the actual loading of various buses carrying both conventional and electric vehicle loads. The deep learning module captures the nonlinear load patterns and flags the potential attacks, while deterministic model validates these attacks by using the voltage data from phasor measurement units (PMUs), thereby minimising the false alarms. It also improves the system resilience and reliability. A three-phase unbalanced radial distribution system (RDS) is considered in this study due to three critical reasons: (a) The charging of EV magnifies the voltage volatility and phase imbalance, making the malicious phase shifts tougher to identify, (b) real-world medium-voltage and low-voltage feeders are inherently unbalanced due to the unequal phase loading, and (c) the asymmetry of an unbalanced system challenges the conventional state estimation-based techniques, necessitating a specialised attack detection strategy. The proposed methodology has been tested on 13-node and 37-node test feeders. Comparative analysis demonstrates how well it protects power distribution networks from such attacks.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A hybrid load redistribution attack detection in unbalanced radial distribution system in presence of EV

  • 期刊论文

  • 发表日期：

    2026年6月22日

查看全文
  • Sayak Mondal,
  • Parimal Acharjee,
  • Aniruddha Bhattacharya

摘要

The rapid integration of electric vehicle charging stations (EVCSs) into the distribution networks introduces the highly unbalanced load patterns which creates the fertile ground for cyber-attacks like load redistribution (LR) attacks. In order to identify load redistribution (LR) attacks, this paper proposes a hybrid approach that combines deep learning and mathematical modelling to ascertain the actual loading of various buses carrying both conventional and electric vehicle loads. The deep learning module captures the nonlinear load patterns and flags the potential attacks, while deterministic model validates these attacks by using the voltage data from phasor measurement units (PMUs), thereby minimising the false alarms. It also improves the system resilience and reliability. A three-phase unbalanced radial distribution system (RDS) is considered in this study due to three critical reasons: (a) The charging of EV magnifies the voltage volatility and phase imbalance, making the malicious phase shifts tougher to identify, (b) real-world medium-voltage and low-voltage feeders are inherently unbalanced due to the unequal phase loading, and (c) the asymmetry of an unbalanced system challenges the conventional state estimation-based techniques, necessitating a specialised attack detection strategy. The proposed methodology has been tested on 13-node and 37-node test feeders. Comparative analysis demonstrates how well it protects power distribution networks from such attacks.