Three-Phase Load Forecasting Model Based on K-Shape and STSGCN

This article proposes a three-phase load forecasting model based on K-shape and Spatial-Temporal Synchronous Graph Convolutional Networks (STSGCN), aiming to improve the accuracy and reliability of three-phase load forecasting. Traditional load forecasting methods struggle to effectively consider the complex global features and correlations among three-phase loads in power systems. This method clusters the loads using K-shape and divides them into nodes based on phase information, then build these categorized nodes into a graph structure, fully considering the correlations between different nodes and use STSGCN model to achieve accurate load prediction. Experimental results on a specific power system's load data demonstrate that the three-phase load prediction method based on K-shape and STSGCN significantly enhances prediction accuracy and stability. This research provides important references and support for the operation and planning of power systems, and offers prospects for future research directions in three-phase load forecasting.