摘要
对碳价波动的特征进行分析,说明碳价预测的意义;然后,基于卷积神经网络(convolutional neural network,CNN)与长短期记忆网络(long short-term memory, LSTM)提出一种CNN-LSTM组合模型的碳价预测方法,充分考虑碳价的时序特性,通过改善相关模型,从时序数据中提取特征的能力从而提高预测准确性;最后,通过欧洲能源交易所及我国广州碳市场的碳价实例验证,将CNN-LSTM模型的预测结果与其他常用预测模型对比,结果表明CNN-LSTM模型在碳价预测中具有更高的预测准确性。
Firstly,it is necessary to analyze the characteristics of carbon price fluctuation and explain the significance of carbon price prediction;Then,based on CNN(convolutional neural network)and LSTM(long short term memory),a CNN-LSTM combined model is proposed to predict the carbon price.This method fully considers temporal characteristics of carbon price and improves the prediction accuracy by improving model's ability to extract features from temporal data.Finally,the prediction results of CNN-LSTM model are compared with other common prediction models by carbon price examples of European Energy Exchange and Guangzhou carbon market.The results show that CNN-LSTM model has higher prediction accuracy in carbon price prediction.
作者
郭宇辰
加鹤萍
余涛
刘敦楠
Guo Yuchen;Jia Heping;Yu Tao;Liu Dunnan(Beijing Key Laboratory of New Energy and Low-carbon Development,Beijing 102206,China;School of Economics and Management,North China Electric Power University,Beijing 102206,China;State Grid Shanghai Municipal Electric Power Company,Shanghai 200122,China)
出处
《科技管理研究》
北大核心
2023年第11期200-206,共7页
Science and Technology Management Research
基金
国家社会科学基金重大项目“面向国家能源安全的智慧能源创新模式与政策协同机制研究”(19ZDA081)。
关键词
碳价预测
长短时记忆网络
卷积神经网络
组合模型
carbon price prediction
long short-term memory network
convolutional neural network
combination model