基于二次分解和RIME-HKELM的中国区域碳价预测  

Predicting Regional Carbon Price in China Based on Secondary Decomposition and RIME-HKELM

作　　者：胡贝贝[1] 程云鹤[1] HU Beibei;CHENG Yunhe(School of Economics and Management,Anhui University of Science&Technology,Huainan,Anhui 232001,China)

机构地区：[1]安徽理工大学经济与管理学院,安徽淮南232001

出　　处：《安徽理工大学学报(社会科学版)》2025年第1期24-35,共12页Journal of Anhui University of Science and Technology:Social Science

基　　金：国家社会科学基金重大项目(22ZDA112);安徽理工大学青年基金项目(QNYB202202)。

摘　　要：碳价有效预测是碳金融市场风险管理的关键。针对中国区域碳价的非线性、非平稳复杂波动特征,提出一种结合二次分解重构策略和霜冰优化算法优化的混合核极限学习机(RIME-HKELM)的组合预测模型,并以湖北碳市场价格为研究对象开展实证研究。结果表明:1)引入RIME算法优化HKELM的参数能改进碳价的预测效果。2)采用基于VMD、改进TVFEMD和极差熵的二次分解重构策略能够提高碳价分解的有效性,进而提高碳价整体的预测性能。3)引入基于RIME-HKELM的非线性集成学习方法确定碳价最终的预测值,能够区分不同子序列对碳价整体预测结果的影响。文中提出的模型相较参照组模型具有显著的预测性能,在碳价预测研究中具有良好的适用性和有效性。Effective prediction of carbon prices matters a great deal for risk management in the carbon financial market.Aiming at the nonlinear and non-stationary complex fluctuations of regional carbon price in China,a hybrid kernel Extreme Learning Machine(RIME-HKELM)combined prediction model optimized by quadratic decomposition reconstruction strategy and frost and ice optimization algorithm was proposed,and the carbon market price in Hubei province was used as the research object for empirical research.The results show that:1)Introducing RIME algorithm to optimize the parameters of HKELM can improve the prediction effect of carbon price.2)The use of a secondary decomposition and reconstruction strategy based on VMD,improved TVFEMD,and range entropy can improve the effectiveness of carbon prices decomposition,thereby enhancing the overall predictive performance of carbon prices.3)Introducing a nonlinear ensemble learning method based on RIME-HKELM to determine the predicted values of residual sequences and the final predicted values of carbon prices can distinguish the impact of different subsequences on the overall prediction results of carbon prices.The model proposed in this article has significant predictive performance compared to the reference group model,and has good applicability and effectiveness in carbon price prediction research.

关 键 词：碳价预测 混合核极限学习机 时变滤波经验模态分解 霜冰优化算法 极差熵 

分 类 号：F803.9[经济管理]