Research on Safety Risk Management of Substation Switching Operations Based on Information Fusion

Substation switching operations are characterized by high frequency and elevated risk levels. Traditional safety prevention methods, which mainly rely on rule-based systems and manual review, often struggle with delayed responses and limited accuracy in complex scenarios. This study proposes a risk identification and early warning method for switching operations based on multi-source information fusion. By integrating multidimensional data such as operation logs, video surveillance, equipment status, and environmental variables, a unified graph structure is constructed. A graph neural network (GNN) combined with a self-attention mechanism is used to extract key risk factors and temporal dependency features. Additionally, a fuzzy analytic hierarchy process (FAHP) is introduced to build a multidimensional risk assessment model, enabling quantitative scoring and hierarchical response to operational risks. Experimental validation on a hardware-in-the-loop simulation platform shows that the proposed method achieves a 91% risk identification accuracy across 300 sample sets, with an average response time of 0.32 seconds, significantly enhancing the intelligent early warning capability and overall safety of switching operations.