基于Stacking集成学习的福建省火电行业碳足迹情景预测

Scenario prediction of carbon footprint for thermal power industry in Fujian Province based on Stacking ensemble learning

提出一种基于Stacking集成学习的福建省火电行业碳足迹情景预测方法.首先,基于扩展Kaya恒等式识别火电行业碳排放的主要影响因素;然后,以决定系数和平均绝对百分比误差作为评价指标优选模型的初级学习器与元学习器,构建融合不同学习器优势的Stacking集成学习碳足迹预测模型;最后,设置4种不同的碳达峰行动情景,以福建省为例对其2021—2035年火电行业的碳达峰进行计算.结果表明,所提方法能够准确预测火电行业的碳足迹,并得出在低碳发展情景下该省可在2027年实现碳达峰战略目标的结论.

A carbon footprint scenario prediction method based on Stacking ensemble learning is proposed.Firstly,the main influencing factors of carbon emissions in thermal power industry are identified based on the extended Kaya constant equation.And then,the decision coefficient and the mean absolute percentage error are selected as the primary learners and meta-learners of the evaluation index model,and the Stacking ensemble learning model for carbon footprint forecasting that integrates the advantages of different learners is established.Finally,four carbon peaking action scenarios are set up to calculate the carbon peaking of thermal power industry in Fujian Province from 2021 to 2035.The results show that the proposed method can accurately predict the carbon footprint of the thermal power industry.It is concluded that the province can achieve the strategic goal of carbon peak in 2027 under the low-carbon development scenario.