基于WOA-ELM模型的中国碳排放预测研究  

Forecast of Carbon Emissions in China Based on WOA-ELM Model

作　　者：王珂珂 牛东晓[1,2] 甄皓 孙丽洁 许晓敏 WANG Keke;NIU Dongxiao;ZHEN Hao;SUN Lijie;XU Xiaomin(School of Economics and Management,North China Electric Power University,Beijing 102206,China;Key Laboratory of New Energy and Low-Carbon Development,Beijing 102206,China)

机构地区：[1]华北电力大学经济与管理学院,北京102206 [2]新能源电力与低碳发展研究中心,北京102206

出　　处：《生态经济》2020年第8期20-27,共8页Ecological Economy

基　　金：教育部哲学社会科学重大课题攻关项目“构建清洁低碳、安全高效的能源体系政策与机制研究”(18JZD032);高等学校学科创新引智计划“中国绿色电力发展研究学科创新引智基地”(B18021);国家自然科学基金资助项目“促进大规模新能源消纳的电力市场管理研究”(71804045)。

摘　　要：为判断中国是否能够实现2030年碳排放强度下降60%~65%的承诺,以及碳排放总量是否能够在2030年达到峰值,论文构建了一个基于鲸鱼优化算法改进的极限学习机模型,对2019-2040年的碳排放量和碳排放强度进行预测。首先基于STIRPAT模型筛选中国碳排放影响因素,将其作为碳排放预测模型的输入因素集,而后运用鲸鱼算法优化极限学习机的参数,克服极限学习机易陷入局部最优解的缺点,最后运用该组合模型对中国未来的碳排放进行预测。通过与ELM、BPNN模型的预测结果和误差指标进行对比,证实论文提出的碳排放预测模型的精度更高,能够更为准确地反映中国未来的碳排放趋势。To determine whether China can achieve its commitment to reduce carbon emissions intensity by 60%-65% in 2030,and whether the total carbon emissions can reach its peak in 2030,this paper construct an improved extreme learning machine model based on whale optimization algorithms to forecast total carbon emissions and carbon emission intensity in 2019-2040.Firstly,China’s carbon emission influencing factors are selected as the input factor set of the carbon emissions forecasting model based on the STIRPAT model.Then,the whale optimization algorithm is used to optimize the parameters of the extreme learning machine,overcome the shortcoming that the extreme learning machine tends to fall into the local optimal solution.Finally,the combined model is used to forecast future carbon emissions of China.By comparison with the forecasting results and error indicators of ELM and BPNN models,it is proved that the carbon emissions forecasting model proposed in this paper has higher accuracy and can more accurately reflect the future carbon emission trends of China.

关 键 词：碳排放 STIRPAT模型 鲸鱼优化算法 极限学习机 

分 类 号：F062.2[经济管理—政治经济学]