Scholarlink

As the integration of smart technologies advances in modern power systems, the vulnerability of smart grids to cyber threats like false data injection becomes a critical concern. This project aimed at enhancing the security of smart grids through the implementation of an advanced false data injection detection mechanism utilizing the isolation forest algorithm. By isolating anomalies in the data, this approach can effectively identify and mitigate malicious data injection attempts that could lead to operational disruptions and compromise the grid’s integrity. The project seeks to develop a robust and efficient solution that not only detects anomalies but also ensures minimal false positives, allowing for timely and accurate responses to potential threats. Through rigorous testing and validation, the proposed approach aims to contribute to the safeguarding of smart grids, bolstering their resilience against evolving cyber threats. Various machine learning algorithms have been used to address this issue, with random forest and XGBoost showing high accuracy scores of 98% and 97%, respectively. Logistic regression, with a lower accuracy of 77%, is useful for binary classification tasks. K-nearest neighbors (KNN) has an 89% accuracy, balancing precision and recall.

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

Identification of False Data Injection Attacks in Smart Grid with Supervised and Ensemble Learning Methods

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Kundu Bhagya Sri,
  • Ch. Raga Madhuri,
  • Kasaraneni Gagana,
  • Tiprineni Sathvika Lakshmi

摘要

As the integration of smart technologies advances in modern power systems, the vulnerability of smart grids to cyber threats like false data injection becomes a critical concern. This project aimed at enhancing the security of smart grids through the implementation of an advanced false data injection detection mechanism utilizing the isolation forest algorithm. By isolating anomalies in the data, this approach can effectively identify and mitigate malicious data injection attempts that could lead to operational disruptions and compromise the grid’s integrity. The project seeks to develop a robust and efficient solution that not only detects anomalies but also ensures minimal false positives, allowing for timely and accurate responses to potential threats. Through rigorous testing and validation, the proposed approach aims to contribute to the safeguarding of smart grids, bolstering their resilience against evolving cyber threats. Various machine learning algorithms have been used to address this issue, with random forest and XGBoost showing high accuracy scores of 98% and 97%, respectively. Logistic regression, with a lower accuracy of 77%, is useful for binary classification tasks. K-nearest neighbors (KNN) has an 89% accuracy, balancing precision and recall.