摘要
随着我国碳排放权交易市场的正式启动,中国碳市场迎来了新的里程碑。准确预测碳配额价格对于数字化社会变革时代下的政策制定和企业决策至关重要。然而,碳配额价格的非稳定性和非线性使得准确预测碳配额市场价格的波动具有一定挑战性。因此,本文构建了一种集可解释性与多尺度分析于一体的差异化学习方法——VMD-AWLSSVR-PSOALS-SHAP混合预测模型。该预测框架考量了影响碳配额价格的潜在因素,同时融入信号分解、高效特征选择、精确价值预测及模型可解释性研究等关键步骤,旨在提高碳配额价格预测的准确性和可理解性,以更好地应对碳市场的复杂性和不确定性。结果表明,该混合预测模型对碳配额价格预测精确度高,对价格影响因素具有可解释性;碳配额价格的影响因素因时间尺度而异,高频序列对短期经济和历史价格敏感,低频序列更容易受到能源价格的影响。
With the official launch of China's carbon emission trading market,China's carbon market has ushered in a new milestone.Accurate forecasting of carbon allowance prices is crucial for policy-making and corporate decision-making in the era of digital social transformation.However,the instability and nonlinearity pose significant challenges for accurate market price forecasting.Therefore,this paper constructs a differentiated learning approach that integrates interpretability and multi-scale analysis,the VMD-AWLSSVR-PSOALS-SHAP hybrid prediction model.The forecasting framework not only comprehensively considers the potential factors affecting the price of carbon allowances,but also incorporates key steps such as signal decomposition,efficient feature selection,accurate value forecasting and model interpretability research,aiming to improve the accuracy and comprehensibility of carbon allowance price forecasting to better cope with the complexity and uncertainty of the carbon market.The results show that the hybrid prediction model achieves high accuracy in predicting the price of carbon allowances and is interpretable to the price influencing factors.The influencing factors of carbon allowance prices vary with time scales,and the high-frequency series are sensitive to short-term economic and historical prices,while the lowfrequency series are more susceptible to the impact of energy prices.
作者
胥杉
胡成春
田浩
刘芮巧
温晓哲
XU Shan;HU Chengchun;TIAN Hao;LIU Ruiqiao;WEN Xiaozhe(Chongqing University of Technology,Banan Chongqing,400054,China)
出处
《对外经贸实务》
2025年第1期40-50,共11页
Practice in Foreign Economic Relations and Trade
基金
国家社科基金项目“中美贸易政策不确定性对我先进制造业全球价值链分工影响与应对研究”(22XJY010)
重庆理工大学研究生创新项目“产业数字化下全球价值链地位对碳排放影响机制研究”(gzlcx20243446)。
关键词
数字化社会变革
机器学习
碳配额价格预测
可解释预测模型
digital social transformation
machine learning
carbon allowance price forecasting
interpretable forecasting model
