In the context of China’s new-type power system construction

this paper proposes an interactive load forecasting framework to address the challenges posed by the increasing complexity of load structure

heightened randomness

and the diverse decision-making needs of the power market. Current load forecasting methods are limited in explainability

adaptability

and support for downstream decision-making

making them inadequate for practical applications in the new-type power system. The proposed interactive framework emphasizes deep interaction between forecasters and algorithms across the stages of data input

model training

and forecast output. It explores four key research directions: enhancing the explainability of complex models

improving model generalization under distribution shift scenarios

integrated modeling to support diverse decision-making in the power market

and designing human-computer interaction mechanisms based on large language models. By reviewing the current research landscape and projecting future research directions

this paper aims to promote the reliable implementation of advanced load forecasting technologies

providing strong technical support for the stable operation of new-type power systems and power grid scheduling.