Zero-shot load forecasting for integrated energy systems: A large language model-based framework with multi-task learning

Jiaheng Li

doi:10.1016/j.neucom.2025.131288

Zero-shot load forecasting for integrated energy systems: A large language model-based framework with multi-task learning

2025-08-19

0

PRE

AI

DOI：10.1016/j.neucom.2025.131288

Original

Original request for help

Save

Abstract

En 中文

• Large Language Models (LLMs) for zero-shot load forecasting. • Semantic space alignment method optimized the generation of prompts for large models. • Multi-task learning mechanism optimized the generation of prompts for multi-feature inputs. • Verified with actual load data from solar-powered households collected by the Australian power grid. • Improved prediction accuracy and transferability.

Keywords:

Large Language Models

zero-shot load forecasting

semantic space alignment

multi-task learning

prompt generation

Zero-shot load forecasting for integrated energy systems: A large language model-based framework with multi-task learning

Abstract

Journal

Organization