基于分形插值与机器学习混合模型的碳价格预测  

Carbon Price Forecasting Based on a Hybrid Model of Fractal Interpolation and Machine Learning

作　　者：何许凡 孙和军[1] 吴锦鸿 李翱宇 HE Xufan;SUN Hejun;WU Jinhong;LI Aoyu(School of Mathematics and Statistics,Nanjing University of Science and Technology,Nanjing 210094,China)

机构地区：[1]南京理工大学数学与统计学院,江苏南京210094

出　　处：《佳木斯大学学报（自然科学版）》2024年第11期172-176,共5页Journal of Jiamusi University：Natural Science Edition

基　　金：江苏省大学生创新创业训练计划项目(202310288193Y)。

摘　　要：碳排放权交易市场是实现碳达峰与碳中和目标的核心工具之一。预测其市场波动对于生产主体稳定预期、推进碳达峰具有重要意义。使用机器学习算法预测插值点,而后延拓原有迭代函数系进行分形插值进行混合预测。将SVM,RF,LSTM分别进行混合,基于广州碳交所数据进行实证分析。结果表明,混合模型较传统插值算法更优,更适用于碳价格这样的非平稳金融时序。The carbon emission trading market is one of the core tools for achieving the goals of carbon peak and carbon neutrality.Predicting its market fluctuations is of great significance for the production entities to have stable expectations and promote carbon peaking.Machine learning algorithms are used to predict interpolation points,and then the original iterative function system is extended for fractal interpolation for hybrid forecasting.SVM,RF,and LSTM are respectively hybridized and empirically analyzed based on the data from the Guangzhou Carbon Exchange.The results show that the hybrid model is superior to traditional interpolation algorithms and is more suitable for non-stationary financial time series such as carbon prices.

关 键 词：碳交易市场 分形插值 机器学习 短期时序预测 

分 类 号：O29[理学—应用数学]