基于Adaboost-SVR模型的我国碳排放强度分析与预测 |
点此下载全文 |
引用本文:梁小林,秦 欢,陈敏茹,许 奇,梁 曌.基于Adaboost-SVR模型的我国碳排放强度分析与预测[J].经济数学,2020,(3):167-174 |
摘要点击次数: 632 |
全文下载次数: 0 |
|
|
中文摘要:首先对我国1960-2017年的碳排放趋势分5个阶段分析,发现虽然在不同时期存在波动,但长期来看,我国碳排放强度呈逐步下降趋势.然后对差分平稳后的序列数据建立Adaboost-SVR预测模型,采用RMSE、MAPE、MAE、MSE四个评价指标比较Adaboost-SVR模型与Adaboost-DT、SVR、BP神经网络对碳排放强度的预测精度.结果表明,组合模型明显优于其他3种模型,对于碳排放强度预测具有很高的可靠性.另外,通过使用Adaboost-SVR模型进行后续年份预测,发现我国未来碳排放强度总体将继续缓慢下降.最后,基于二氧化碳排放量的LMID分解结果,提出调整能源产业结构, 促进可再生能源利用等节能减排建议. |
中文关键词:碳排放强度 自适应提升算法 支持向量回归 LMDI分解 |
|
Analysis and Prediction of China's Carbon Emission Intensity Based on Adaboost-SVR Model |
|
|
Abstract:The trend of China's carbon emission from 1960 to 2017 is analyzed in five stages. It is found that the intensity of China's carbon emission is gradually decreasing in the long run, although there are fluctuations in different periods. Then Adaboost-SVR model is established on the basis of differential data. Four evaluation indexes including RMSE,MAPE,MAE,MSE are used to compare the prediction accuracy of Adaboost-SVR and Adaboost-DT, SVR, BP neural network for carbon emission intensity. The results show that the combined model is superior to the other three models, and has a high reliability for carbon emission intensity prediction. The Adaboost-SVR model is used to predict the subsequent years, and the prediction results show that China's carbon emission intensity will continue to gradually decline in the future. Finally, based on the results of LMDI decomposition of carbon dioxide emissions, some suggestions are put forward, such as adjusting the energy industrial structure and promoting the use of renewable energy. |
keywords:carbon emission intensity adaboost SVR LMDI decomposition |
查看全文 查看/发表评论 下载pdf阅读器 |