工业技术经济2025，Vol.44Issue(7)：124-134,11.DOI:10.3969/j.issn.1004-910X.2025.07.012
基于机器学习Mod-EMD-BiLSTM组合模型的碳价格预测方法研究
Research on Carbon Price Prediction Method Based on Machine Learning Mod-EMD BiLSTM Combination Model
摘要
Abstract
Accurate prediction of carbon prices can provide quantitative support and reference basis for climate policy formu-lation,rational decision-making of investors,and maintaining the stable operation of the carbon market.This article proposes a no-vel hybrid machine learning Mod-EMD-BiLSTM prediction model that combines data augmentation,empirical mode decomposition,and bidirectional long short-term memory techniques.Specifically,the original carbon price series is first subjected to empirical mode decomposition to obtain a series of relatively stable and low noise intrinsic mode components(IMF).Secondly,introduce da-ta augmentation techniques to enhance data recombination and randomly generate half of the IMF combinations.Furthermore,based on the two parallel mechanisms of prevention and prediction in this model,the IMF composite components are further preprocessed and model training is carried out.Finally,the output values of the two frameworks are integrated through the fully connected layer of the BiLSTM neural network to obtain the final carbon price prediction results.On the basis of establishing a prediction model,em-pirical research is conducted by crawling the daily closing prices of carbon trading in Hubei's carbon trading market from 2014 to 2024.The results show that the model established in this article exhibits the best direction prediction accuracy compared to the other 16 benchmark models,reflecting the superior prediction performance and good practicality of the model.关键词
碳价格预测/机器学习/经验模态分解/BiLSTM/数据增强/碳市场机制/方向预测精度/时间序列分析Key words
carbon finance/machine learning/empirical mode decomposition/BiLSTM/data augmentation technology/carbon market mechanism/directional prediction accuracy/time series analysis分类
经济学引用本文复制引用
赵玉帛,丁晓格,刘露毅,张旭晴..基于机器学习Mod-EMD-BiLSTM组合模型的碳价格预测方法研究[J].工业技术经济,2025,44(7):124-134,11.基金项目
河南省哲学社会科学规划项目"河南数字经济产业创新生态构建与治理研究"(项目编号:2023CJJ113) （项目编号:2023CJJ113）
河南省软科学项目"数实产业技术融合赋能河南制造业企业创新韧性的机制与提升策略研究"(项目编号:252400410090). （项目编号:252400410090）
