我国碳排放增长率的运行机理及预测  

The Running Mechanism and Prediction of the Growth Rate of China's Carbon Emissions

作　　者：张国兴[1] 张振华[1] 刘鹏[2] 刘明星[1] 

机构地区：[1]兰州大学管理学院,甘肃兰州730000 [2]中国科学院数学与系统科学研究院,北京100080

出　　处：《中国管理科学》2015年第12期86-93,共8页Chinese Journal of Management Science

基　　金：国家自然科学基金资助项目(71103077);教育部新世纪优秀人才支持计划项目(NCET-13-0267);教育部人文社会科学基金项目(15YJA630097);兰州大学中央高校基本科研业务费项目(15LZUJBWYJ040)

摘　　要：碳排放是气候变暖的重要原因之一,研究和预测碳排放增长率能为低碳政策的制定提供理论指导。利用经验模态分解方法,本文将我国碳排放增长率序列分解为短期波动项和趋势项两个序列,并分析了国家政策、国内宏观经济变化、金融危机对短期波动项和趋势项的影响。在此基础上,利用动态神经网络分别对趋势项和短期波动项进行预测,并将二者之和作为最终的碳排放增长率的预测值。最后,从误差序列绝对值的最大值、最小值、均值和标准差四个角度来比较该预测方法与单独以碳排放量和碳排放增长率为输入变量的神经网络模型的优劣,并得出本文提出的模型具有预测有效性的结论。It can provide theoretical guidance for the development of low-carbon policies to research and forecast the growth rate of carbon emissions since carbon emissions have become an important reason for global warming. The growth rate of China＇s carbon emissions was decomposed into short-term volatility and trend term these two sequences by the use of empirical mode decomposition method, and the national policy, the domestic macroeconomic changes and the financial crisis influence to the short-term volatility and trend term were analyzed respectively. On this basis, using dynamic neural network to forecast the short-term volatility and trend term, and sum the two predicted values as the final growth rate of carbon emissions. Finally, from the error sequence of absolute value maximum, minimum, mean and standard deviation of the four angles to compare this prediction method with neural network model that separate input variables of carbon emissions or the growth rate of carbon emission, and found that the model presented in this paper can predict the growth rate effectively.

关 键 词：碳排放增长率 经验模态分解 预测 神经网络 

分 类 号：F201[经济管理—国民经济]