Research on Electromagnetic Interference Tracing and Multi-Scale Protection Optimization in Substations Based on Transformer

This paper presents a Transformer-based model for electromagnetic interference (EMI) source localization and multi-scale protection optimization in substations. As modern substations are increasingly subjected to various sources of EMI, including switching transients, voltage spikes, harmonics, and lightning surges, their protection becomes more complex. Traditional mitigation strategies often fail to address the dynamic and multi-source nature of these disturbances. To overcome this challenge, the paper proposes the use of Transformer deep learning models, which are capable of handling sequential data and capturing long-range dependencies in time-series signals. The proposed Transformer model localizes and classifies different types of interference, offering real-time detection of disturbance sources within the substation. Furthermore, a multi-scale protection strategy is introduced, combining physical, electrical, and operational protection measures to minimize the impact of EMI. The effectiveness of this approach is validated through simulation-based experiments, demonstrating significant improvements in localization accuracy and interference suppression when compared to traditional methods. This research offers a comprehensive, intelligent solution for enhancing the resilience of substations against EMI, contributing to the development of smarter, more reliable power grids.