STAAGCN and Preference-Guided NSGA-III for Transmission Expansion Planning

The rapid growth of electricity demand poses significant challenges to medium and long term grid planning. Traditional methods, relying on coarse growth rates or typical daily curves, typically lack mechanisms to incorporate engineering preferences into multi-objective optimization. To address this, this paper proposes a coordinated framework integrating an adaptive spatiotemporal graph attention network with a preference-guided multi-objective optimizer. The former captures dynamic spatial couplings for high-accuracy load forecasting, while the latter enhances NSGA-III via a weighted priority survival strategy to jointly minimize investment, operation, and load-shedding costs. Validated on the IEEE 30-bus transmission test system, the framework demonstrates strong decision robustness under uncertainty, significantly improving forecasting accuracy and computational efficiency while reducing total planning costs. This work provides an intelligent, end-to-end decision-support tool explicitly designed for medium to long-term transmission expansion planning.