Department of Economics and Management, North China Electric Power University, Baoding, Hebei, China.
Department of Economics and Management, North China Electric Power University, Baoding, Hebei, China.
Sci Total Environ. 2020 May 20;718:137194. doi: 10.1016/j.scitotenv.2020.137194. Epub 2020 Feb 13.
In recent years, with the worsening of the global climate problem, the issue of CO emissions has gradually attracted people's attention. Accurately predicting CO emissions and analyzing its change trends are important elements in addressing climate issues at this stage. Although the predecessors have done a lot of research on CO emissions and also established some prediction models, few people have adopted quantifiable methods to select prediction indicators and studied the CO emissions of commercial department. So this paper establishes a novel BP neural network prediction model based on the index quantization ability of random forest and the performance optimization ability of PSO. For further strengthening the prediction accuracy, several improvements have been made to PSO. Finally, the validity of the model is tested using panel data from 1997 to 2017 of the Chinese commercial sector. The results as follows: (1) Compared with other parallel models, the newly established hybrid forecasting model can more accurately predict the CO emissions of China's commercial department. (2) The prediction indexes selected after quantification based on the random forest can improve the prediction accuracy. (3) These improvements of PSO in this paper can greatly enhance the prediction effect of the hybrid prediction model.
近年来，随着全球气候问题的恶化，CO2 排放问题逐渐引起人们的关注。准确预测 CO2 排放并分析其变化趋势是现阶段解决气候问题的重要因素。尽管前人对 CO2 排放做了大量的研究，并建立了一些预测模型，但很少有人采用可量化的方法来选择预测指标，并研究商业部门的 CO2 排放。因此，本文基于随机森林的指标量化能力和 PSO 的性能优化能力，建立了一种新的 BP 神经网络预测模型。为了进一步提高预测精度，对 PSO 进行了一些改进。最后，利用中国商业部门 1997 年至 2017 年的面板数据对模型的有效性进行了验证。结果表明：（1）与其他并行模型相比，新建立的混合预测模型可以更准确地预测中国商业部门的 CO2 排放。（2）基于随机森林进行量化选择的预测指标可以提高预测精度。（3）本文对 PSO 的这些改进可以极大地增强混合预测模型的预测效果。
