Scholarlink

<p>Application programming interfaces (APIs) form the backbone of modern power system digitalization, enabling real-time data exchange, remote control, and cross-domain orchestration across SCADA, substation automation, and dispatching systems. However, as the digital grid evolves toward highly interconnected architectures, these APIs have become increasingly exposed to adversarial manipulation and stealthy misuse. Their reliability directly impacts the stability and resilience of grid operation and long-term cybersecurity assurance. Existing anomaly detection approaches in power systems often rely on statistical correlations or isolated secure aggregation mechanisms, which face two key challenges: (i) high sensitivity to benign operational drifts, leading to spurious false alarms under changing load and communication patterns; and (ii) lack of guarantees, resulting in fabricated or un anomaly reports under adversarial interference. To address these issues, we propose CausalVeri, a causality-aware verification framework that integrates causal representation learning with a verifiably secure aggregation protocol. Experiments demonstrate that CausalVeri reduces false positives by up to 17%, maintains stable detection accuracy under non-stationary grid communication patterns, and ensures cryptographic verifiability of anomaly alerts, providing a auditable foundation for secure API-based interactions in digital power systems.</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

CausalVeri: causal-aware and cryptographically anomaly detection for power-system APIs

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月14日

查看全文
  • Xuhua Ai,
  • Yun Dong,
  • Yuan Yin,
  • Yiting Huang,
  • Qiwen Tan,
  • Zijian Lin

摘要

Application programming interfaces (APIs) form the backbone of modern power system digitalization, enabling real-time data exchange, remote control, and cross-domain orchestration across SCADA, substation automation, and dispatching systems. However, as the digital grid evolves toward highly interconnected architectures, these APIs have become increasingly exposed to adversarial manipulation and stealthy misuse. Their reliability directly impacts the stability and resilience of grid operation and long-term cybersecurity assurance. Existing anomaly detection approaches in power systems often rely on statistical correlations or isolated secure aggregation mechanisms, which face two key challenges: (i) high sensitivity to benign operational drifts, leading to spurious false alarms under changing load and communication patterns; and (ii) lack of guarantees, resulting in fabricated or un anomaly reports under adversarial interference. To address these issues, we propose CausalVeri, a causality-aware verification framework that integrates causal representation learning with a verifiably secure aggregation protocol. Experiments demonstrate that CausalVeri reduces false positives by up to 17%, maintains stable detection accuracy under non-stationary grid communication patterns, and ensures cryptographic verifiability of anomaly alerts, providing a auditable foundation for secure API-based interactions in digital power systems.