Scholarlink

Purpose <p>A new Temporal Inductive Path Neural Network (TIPNN) for sophisticated defect detection in power transmission lines is presented in this paper.</p> Methods <p>TIPNN models both temporal dynamics and inductive signal pathways using a hybrid deep learning architecture combining temporal convolutional networks and inductive graph learning. Under dynamic load and climatic settings, experimental evaluations were conducted on a simulated smart grid dataset enhanced with different fault types, such as line-to-ground, line-to-line, and three-phase faults.</p> Results <p>Quantitative results highlight the TIPNN's superior performance, achieving a fault classification accuracy F1-score of 98.57%, recall of 98.53%, precision of 98.61%, and 98.74%. Additionally, the model demonstrated robust generalization across different noise levels, with only a 1.4% drop in accuracy under 10% Gaussian noise. Compared to traditional models showed a performance improvement of 4.8%–7.2% in accuracy and reduced detection latency by over 35%.</p> Conclusion <p>These results substantiate the TIPNN as a scalable and accurate solution to diagnose problems in smart power grids in actual time.</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

Advanced Power Transmission Line Fault Recognition in IoT-Based Smart Grids Using Temporal Inductive Path Neural Networks

  • 期刊论文

  • 发表日期：

    2025年11月28日

查看全文
  • R. Karthick,
  • M. Mathivanan,
  • T. Anil Kumar,
  • Koganti Srilakshmi

摘要

Purpose

A new Temporal Inductive Path Neural Network (TIPNN) for sophisticated defect detection in power transmission lines is presented in this paper.

Methods

TIPNN models both temporal dynamics and inductive signal pathways using a hybrid deep learning architecture combining temporal convolutional networks and inductive graph learning. Under dynamic load and climatic settings, experimental evaluations were conducted on a simulated smart grid dataset enhanced with different fault types, such as line-to-ground, line-to-line, and three-phase faults.

Results

Quantitative results highlight the TIPNN's superior performance, achieving a fault classification accuracy F1-score of 98.57%, recall of 98.53%, precision of 98.61%, and 98.74%. Additionally, the model demonstrated robust generalization across different noise levels, with only a 1.4% drop in accuracy under 10% Gaussian noise. Compared to traditional models showed a performance improvement of 4.8%–7.2% in accuracy and reduced detection latency by over 35%.

Conclusion

These results substantiate the TIPNN as a scalable and accurate solution to diagnose problems in smart power grids in actual time.