计算机与现代化Issue(7)：55-62,8.DOI:10.3969/j.issn.1006-2475.2025.07.008
基于优化Transformer的长短期空气污染物浓度预测
Long-and Short-Term Air Pollutant Concentration Forecasting Based on Optimized Transformer
摘要
Abstract
Addressing the issues of low prediction accuracy,short timeliness,and difficulties in capturing spatiotemporal fea-tures for air pollutant concentration prediction,a Transformer architecture based on conditional mask self-attention is proposed,named CondMSA-Transformer.This paper improves the multi-head self-attention mechanism in the Transformer model,intro-duces the sparse attention concepts.By integrating critical environmental factors such as wind speed and wind direction,it imple-ments intelligent"masking"of unnecessary site data,focusing on extracting the most valuable information within the spatiotem-poral dimension.This strategy effectively avoids interference from weak signals of remote stations,reduces computational com-plexity,and enhances the model's ability to capture core features.Comprehensive experimental evaluations on two real datasets in Beijing demonstrate that CondMSA-Transformer exhibits robust performance in both short-term and long-term prediction sce-narios,providing up to 14.67%improvement in mean absolute error(MAE)for PM2.5 prediction compared to other existing meth-ods.This shows its vast application potential and advancement in the field of air quality prediction.关键词
空气污染物浓度预测/长期预测/Transformer/注意力机制/条件掩码Key words
air pollutant concentration prediction/long-term predictions/Transformer/attention mechanism/conditional mask分类
资源环境引用本文复制引用
蔡博涵,刘俊..基于优化Transformer的长短期空气污染物浓度预测[J].计算机与现代化,2025,(7):55-62,8.基金项目
国家自然科学基金面上项目(62273126) （62273126）
