摘要
根据萧山1997—2017年碳排放数据和"十四五"规划指标,基于ARIMA时间序列模型、NAR神经网络、STIRPAT模型分别预测2025年萧山碳排放量.运用最优加权组合模型,截取模型2007—2016年的拟合数据,将3种模型进行组合预测.结果显示,组合之后各项评价指标均表明相对单一模型更好,组合预测模型的拟合优度为0.83,相对平均绝对误差为3.14%、均方根误差为0.5643,组合模型有更高的精度.组合模型预测2025年萧山的碳排放量将达到2454.98万t,碳排放强度下降至0.817 t/万元.
According to the carbon emission data of Xiaoshan from 1997 to 2017 and the index of the 14th Five-Year Plan,the carbon emission of Xiaoshan in 2025 is predicted based on ARIMA time series model,NAR neural network and STIRPAT model.The optimal weighted combination model was used the fitting data of the model from 2007 to 2016 were intercepted to make the combination prediction.The results showed that the evaluation indexes after the combination were better than the single model,the goodness of fit of the combination prediction model was 0.83,the relative mean absolute error was 3.14%,and the root mean square error was 0.5643,and the combination forecasting model had higher accuracy.The combination forecasting model predicts that the carbon emission of Xiaoshan will reach 24.5498 million tons in 2025,and the carbon emission intensity will decrease to 0.817 tons/ten thousand yuan.
作者
罗曼
余彬
翁利国
徐源
龙妍
LUO Man;YU Bin;WENG Li-guo;XU Yuan;LONG Yan
出处
《节能》
2022年第4期75-80,共6页
Energy Conservation
