Multi-task learning for solving OPF in an evolving environment

doi:10.1016/j.apenergy.2025.127174

Multi-task learning for solving OPF in an evolving environment

2025-12-04

0

PRE

AI

DOI：10.1016/j.apenergy.2025.127174

Original

Original request for help

Save

Abstract

En 中文

• To tackle dataset imbalance challenge: we propose an error-focused up-sampling method to balance training samples for each scenario and avoid biased training. • To tackle weak correlation challenge: we propose a multi-task learning framework to achieve knowledge transfer across four scenarios. • To further balance the training process: we adopt an adaptive-weight algorithm to balance the training process. This is achieved by applying trainable weights to each training sample individually.

Multi-task learning for solving OPF in an evolving environment

Abstract

Journal

Organization