Prediction Of Carbon Emissions Based On Modal Decomposition And Neural Network Model

The increase in carbon dioxide emissions and the associated environmental pressures has caused worldwide concern.It is a direct contributor to global warming,the greenhouse effect and extreme weather（including frequent hurricanes,flooding disasters and high temperatures）.The environment has been severely damaged by global warming,melting glaciers,and annual sea level rise.In addition,tense international relations and local conflicts have become more frequent.The rebound of carbon emissions in the past two years has alerted us that reducing carbon emissions is a long-term process,and setting carbon emission reduction targets and policies can be more effective in curbing the rising trend of carbon emissions,so it is especially important to conduct research on carbon emission forecasting.This paper applies the research idea of decomposition before prediction to carbon emission prediction.Based on the EEMD and the PSOBP,a decomposition forecasting combination model is constructed to predict the daily monitoring data of carbon emissions in China,and EEMD has been applied to the field of carbon emission prediction.In this paper,the prediction of carbon emissions is accurate to monthly or quarterly,which is helpful for policy makers to be more alert to the changing trend of carbon emissions and dynamically adjust the target of carbon emission reduction.The research of this paper provides theoretical support for the formulation of carbon emission reduction targets and the adoption of carbon emission reduction measures,and provides reference for the prediction of other greenhouse gases,which is of great significance for achieving carbon emission reduction targets more effectively and quickly.Based on EEMD and PSO-BP,this paper constructed a combined model to predict carbon emissions.Taking the daily monitoring data of carbon emissions in China from January 1,2019 to September 30,2020 as an example,14 comparative models are constructed in this paper.The prediction effect of these model was evaluated by three evaluation indexes,²,RMSE and MAPE.The results showed that the prediction indexes have been greatly improved after decomposition of EEMD,EMD,FEEMD and CEEMD,and the improvement of all evaluation indexes of EEMD is more than 50%.The results show that EEMD-PSOBP model has the best prediction performance among all model combinations.After verifying the superiority of the model,this paper predicts the carbon emissions from March,2022 to July 2022,and the change trend of carbon emission in the future is obtained.The carbon emission in March,April,May,June and July is 839.68,790.4336,823.5175,838.2573 and 878.2325₂/month respectively.Based on the research of this paper,the following conclusions can be drawn:（1）Data decomposition can significantly improve the accuracy of carbon emission prediction;（2）EEMD decomposition can significantly improve the prediction performance of the model;（3）according to the evaluation criteria used in this paper,the prediction effect of EEMD-PSO-BP model is superior.Therefore,this model has a strong potential to be popularized in the field of carbon emission prediction to achieve higher accuracy.