Year: 2025
Type: conference-paper
Abstract: China plans to peak carbon emissions in 2030 and achieve carbon neutrality by 2060. It is important to forecast CO₂ concentration trends accurately for carbon reduction policy-making and promoting sustainable development. In this paper, a deep learning model, AC-BiLSTM, which combines CNN and BiLSTM with an Attention mechanism, is proposed to improve hourly CO₂ concentration forecasting. Features are extracted locally by CNN, bidirectional patterns are encoded by BiLSTM, and Attention helps to focus on informative information for the purpose of supporting accuracy. Experiments show that AC-BiLSTM improves predictions of 24-hour CO₂ levels and performs well on intricate time-series datasets.
Authors: Yue Jia, Shanliang Pan
Institution: Ningbo University
Language: English
Subfield: Environmental Engineering
Field: Environmental Science
Domain: Physical Sciences
SDG: Climate action
Open Access status: hybrid
OpenAlex ID: W4413812503