Securing Energy Internet Construction: An Optical Fiber Fence Intrusion Monitoring Method Based on Improved Spatial Attention Mechanism

Given the complexity and various potential risks present at construction sites of energy internet, particularly the issue of illegal intrusions during the suspension phase of energy internet construction, this paper proposes an optical fiber fence intrusion monitoring method based on an improved spatial attention mechanism. Our method is based on a widely used optical fiber fence intrusion monitoring systems across various industrial sectors. First, endpoint detection is applied to the vibration signals, and noise is removed before transforming the one-dimensional sequences into two-dimensional graphs using Gramian Angular Field (GAF) encoding. Then, an improved Spatial Attention Mechanism (ISAM) module is added to the ResNet50 network, combined with Binary Cross-Entropy (BCE) Loss and Center Loss to enhance classification performance. Test results show that the proposed method achieves a detection accuracy of 98.58% for illegal intrusion detection during the suspension phase of energy internet construction, significantly outperforming traditional methods in terms of both accuracy and real-time, while reducing the false alarm rate and missed detection rates.