文献类型：会议论文

英文题名：Carbon Price Forecasting for China's Eight Major Markets Based on GRU-Attention Model

作者：Wang, Mingcheng[1];Hu, Qingchun[1];Zhu, Wei[1];Huang, Jiani[1]

机构：[1]East China Univ Sci & Technol, Shanghai, Peoples R China

会议论文集：2nd Asia Conference on Computer Vision, Image Processing and Pattern Recognition (CVIPPR)

会议日期：APR 26-28, 2024

会议地点：Xiamen, PEOPLES R CHINA

语种：英文

外文关键词：GRU-Attention; Carbon price forecasting; Time series

摘要：Carbon price can not only be used to judge environmental quality, but also assist in policy designation and structural transformation. However, predicting carbon price is still a difficult task due to the high volatility of carbon price. At this stage, carbon price prediction models mainly include econometric models and machine learning models. The former can only capture the linear features while the latter is challenging to predict the long time series. The paper proposes a carbon price prediction model based on GRU-Attention network. By introducing the attention mechanism on the basis of GRU network, the sensing field is expanded to capture the temporal features of long sequences. The model is experimented on eight major carbon trading markets in China. Experimental results show that the average performance on the three error indicators of MAE, MAPE and RMSE is 1.1697%, 3.4562% and 1.7881%, respectively. The proposed model outperforms other benchmark models in the majority of markets, indicating the effectiveness of the model.