Scholarlink

<p>In this paper, we present a distributed fibre optic sensing technology and deep learning-based safety monitoring system for construction work in distribution networks. First, we suggest a technique that makes use of quasi-distributed intelligent fibre optic sensing technology and distributed fibre optic temperature sensing technology to monitor the operational environment of power distribution networks. This technique can foresee possible safety risks and monitor variables like temperature and stress in real time. Second, we use a k-means based SMOTE data enhancement technique to expand the data for minority class events in order to improve the performance of the deep learning model. This addresses the issues of data imbalance and difficulties in recognizing minority class events. Next, in order to efficiently identify security concerns in construction sites, we suggest a CNN-LSTM network-based approach for security event recognition. This method combines convolutional neural networks with long short-term memory networks. Ultimately, modelling studies and field tests yield positive results, confirming the system’s viability and efficacy and offering a fresh approach to enhancing the security of construction projects in power distribution networks.</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

Key technologies and implementation of vehicle mounted system for distribution network inspection based on laser and visual SLAM

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月18日

查看全文
  • Lichao Cui,
  • Ying Xie,
  • Tianjia Qiu,
  • Yongjian Zhang,
  • Yi Jiang

摘要

In this paper, we present a distributed fibre optic sensing technology and deep learning-based safety monitoring system for construction work in distribution networks. First, we suggest a technique that makes use of quasi-distributed intelligent fibre optic sensing technology and distributed fibre optic temperature sensing technology to monitor the operational environment of power distribution networks. This technique can foresee possible safety risks and monitor variables like temperature and stress in real time. Second, we use a k-means based SMOTE data enhancement technique to expand the data for minority class events in order to improve the performance of the deep learning model. This addresses the issues of data imbalance and difficulties in recognizing minority class events. Next, in order to efficiently identify security concerns in construction sites, we suggest a CNN-LSTM network-based approach for security event recognition. This method combines convolutional neural networks with long short-term memory networks. Ultimately, modelling studies and field tests yield positive results, confirming the system’s viability and efficacy and offering a fresh approach to enhancing the security of construction projects in power distribution networks.