Knowledge-Data Integrated Conformal Transformer for Short-Term Bus Load Forecasting

View at publisher Summarise with VIRAT
Authors: Bin Feng, Gang Huang, Wei Gu, Wenda Zhan, Wenjuan Mei, Chuangxin Guo
Cited by: 0
References: 14

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Considering the challenges posed by the integration of renewable and distributed energy sources into power grids, this paper proposes a short-term bus load forecasting framework that integrates electric-power domain knowledge and data-driven approaches. Firstly, a feature engineering system comprising historical load features, meteorological features, comfort index features, and time–electricity price combined features is established to effectively capture load variation patterns. Secondly, a Transformer-based quantile regression and conformal prediction model is developed to enhance forecasting accuracy and quantify uncertainty. Finally, systematic model comparison and feature ablation experiments are conducted using actual bus load data from Zhejiang Province, China, verifying the effectiveness and engineering applicability of the proposed method. Experimental results demonstrate that the proposed Conformal Transformer method significantly outperforms traditional MLP and standard Transformer models in terms of prediction accuracy and prediction interval reliability. This research provides important support for achieving more precise and reliable decision-making in power system operation.

Fields of Study & Subjects

Energy Load and Power Forecasting Traffic Prediction and Management Techniques Smart Grid and Power Systems

Publication details

Journal / source
IEEE PES International Meeting (PES IM)
Pages
1–5
Publication date
18 January 2026
Language
EN