Scholarlink

<p>Power service process monitoring faces critical challenges in capturing complex spatiotemporal dependencies and identifying anomalies across distributed operational networks. This paper proposes an intelligent monitoring system incorporating spatiotemporal attention mechanisms to address these limitations. The system features a hierarchical attention architecture that jointly models temporal evolution patterns within service workflows and spatial correlations across regional centers, coupled with an adaptive threshold mechanism for anomaly detection. Experimental validation using real-world data from multiple power utilities demonstrates superior performance, achieving 96.84% accuracy and 96.0% recall in field deployment. The system reduces average process completion time by 20.3% and customer complaints by 31.2% across 32 service centers during a 6-month trial. Results confirm that explicit joint spatiotemporal modeling significantly outperforms conventional approaches, providing actionable insights for proactive process optimization in power utility operations.</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

Intelligent monitoring and anomaly detection for power service processes based on spatiotemporal attention mechanism

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月7日

查看全文
  • Nvgui Lin,
  • Xiaobin Wen,
  • Jiacheng Wu,
  • Xiaoying Huang,
  • Hanfei Wen

摘要

Power service process monitoring faces critical challenges in capturing complex spatiotemporal dependencies and identifying anomalies across distributed operational networks. This paper proposes an intelligent monitoring system incorporating spatiotemporal attention mechanisms to address these limitations. The system features a hierarchical attention architecture that jointly models temporal evolution patterns within service workflows and spatial correlations across regional centers, coupled with an adaptive threshold mechanism for anomaly detection. Experimental validation using real-world data from multiple power utilities demonstrates superior performance, achieving 96.84% accuracy and 96.0% recall in field deployment. The system reduces average process completion time by 20.3% and customer complaints by 31.2% across 32 service centers during a 6-month trial. Results confirm that explicit joint spatiotemporal modeling significantly outperforms conventional approaches, providing actionable insights for proactive process optimization in power utility operations.