摘要
新兴的中国区域性碳排放市场受到交易制度,异质环境,政策等因素的影响,使碳价呈现出非线性,非平稳,多频率等特征,风险较为突出.研究碳价的预测方法,有利于碳市场风险管理.传统的单一模型不能全面刻画碳价波动特征,论文构建多频率组合预测模型.运用极点对称模态分解方法将碳价时间序列分解为互不耦合的模态分量;将这些分量分为高,中,低频部分,分别选择适合三种不同频率模态下的预测方法NAR(non-linear autoregressive),WNN(wavelet neural network),SVM(support vector machine)确定其输入输出结构以分类预测;利用PSO-SVM集成碳价分类预测结果,发现:与NAR,WNN,SVM,GARCH等单模型相比,论文的多频率组合预测模型精度更高,是一种更为有效的碳价预测方法.
China's regional carbon emissions trading price is a nonlinear, non-stationary, and multi- frequency time series due to trading system, heterogeneous environment, and policy. With the highlighted risk of carbon emissions trading, the method research of carbon price forecast is important for carbon market risk management. In order to solve the problem that single model cannot fully describe the characteristics of carbon price fluctuation, this paper establishes a multi-frequency combination forecast model. First, extreme-point symmetric mode decomposition (ESMD) is used to decompose the original nonlinear and non-stationary carbon price series into several uncoupling intrinsic mode functions (IMFs). Then, IMFs are divided into three different frequencies: high, medium, and low frequency. Third, non-linear autoregressive (NAR), wavelet neural network (WNN), and support vector machine (SVM) are adopted to forecast each frequency data. Finally, PSO-SVM is used to integrate predicted results. Compared with NAR, WNN, SVM, and GARCH models, the empirical results show that the multi-frequency composite model can improve the model accuracy. Besides, the proposed integrated model proves more efficient and consistent.
作者
张晨
杨仙子
ZHANG Chen YANG Xianzi(School of Management, Hefei University of Technology, Hefei 230009, China Key Laboratory of Process Optimization and Intelligent Decision-making, Hefei University of Technology, Hefei 230009, China)
出处
《系统工程理论与实践》
EI
CSSCI
CSCD
北大核心
2016年第12期3017-3025,共9页
Systems Engineering-Theory & Practice
基金
国家自然科学基金(71373065)~~
关键词
区域性碳排放交易市场
碳价预测
极点对称模态分解
多频率组合预测
regional carbon emission trading market
carbon price forecasting
extreme-point symmetric mode decomposition
multiple frequency combination forecast