Carbon price prediction model based on adaptive variational mode decomposition and optimized extreme learning machine

The open carbon trading market is an important means to reduce carbon emissions, develop a low-carbon economy, and promote environmental protection. Accurate carbon price projections have far-reaching implications for economic and environmental policymaking. In this paper, a new combination model of carbon price prediction is proposed, which can effectively improve the accuracy of prediction. First, adaptive variational mode decomposition (AVMD) was used to decompose the original carbon price sequence into several less complex intrinsic mode functions (IMF). Secondly, the partial autocorrelation function (PACF) is used to determine the input variables for each IMF forecasting model. Finally, the extreme learning machine (ELM) optimized by the dynamic adaptive inertial factor particle swarm optimization (DAIFPSO) algorithm was used for prediction. The proposed combination model is applied to the carbon price prediction of some carbon markets in China (Beijing, Guangdong, Hubei) and compared with the other 11 models. Finally, it is found that the model has the best performance. The results show that the model is effective and stable. Therefore, this model can be used as an effective tool for government and carbon market participants to predict carbon prices and provide a reference for their decision-making.