A multi-factor forecasting model for carbon emissions based on decomposition and swarm intelligence optimization

With the growth of national economy, carbon emissions are increasing day by day, which has caused a series of adverse effects on people's production, life and environment. Therefore, it is very important to establish an accurate carbon emissions forecasting system for developing a low-carbon economy. Aiming at the problems of many influencing factors and randomness of carbon emissions, a multi-factor forecasting model for carbon emissions with generalized phase permutation entropy (GPPE), singular spectrum decomposition (SSD), bidirectional long short-term memory (BiLSTM) optimized by artificial tasmanian devil optimizer (TDO), named GPPE-SSD-TDO-BiLSTM, is proposed. Firstly, Pearson correlation coefficient method is used to analyze the main influencing factors of carbon emissions. Secondly, carbon emissions is decomposed by GPPE-SSD to obtain different singular spectrum components (SSCs). Finally, each component and influencing factors are brought into the TDO-BiLSTM forecasting model, and each forecasting result is reconstructed to obtain the final forecasting result. To verify the validity of the proposed model, carbon emissions of China and India in recent five years are used as experimental data, compares eight different forecasting models, and evaluates the forecasting models with six evaluation indexes. Taking China's carbon emissions as an example, the MAE, MAPE, RMSE, R2 and STD of the proposed model can reach 0.1593, 0.5226, 0.2131, 0.9973 and 3.1056 respectively, which shows that its forecasting effect is better than that of other comparative models.

Automated summary

This study proposes a multi-factor forecasting model for carbon emissions, considering the various influencing factors and randomness of carbon emissions. The model decomposes carbon emissions using GPPE-SSD and applies TDO-BiLSTM for forecasting. The results show that the proposed model outperforms other comparative models in terms of forecasting accuracy.