Research On Carbon Price Prediction Based On Error Decomposition Correction Model

Due to the impact of the development trend of economic globalization,various extreme ecological events caused by global warming make carbon emission reduction urgent.Among them,the carbon emission trading system based on market trading mechanism has been proved to be an effective emission reduction tool.Therefore,in order to provide an effective way to achieve the carbon emission reduction target,China successively launched eight national carbon emission trading rights markets in 2017 to accumulate practical experience for promoting the establishment of a unified national carbon emission trading system.The carbon emission trading price,as a key factor in the carbon trading market,can reflect the market supply and demand and operation,and help enterprise investors and traders realize effective trading strategies in the carbon market.Therefore,accurate carbon price prediction is an important link in carbon emission reduction.Based on this,this paper predicts carbon prices by constructing an error separation correction prediction model,and improves prediction accuracy on the basis of a combined model,and broadens the research ideas for carbon prices.Specifically,the error decomposition method is introduced into the model,combined with artificial intelligence algorithms such as complementary set empirical decomposition(CEEMD),least squares support vector machine(LSSVM),variational modal decomposition(VMD)and extreme value learning machine(ELM),A CEEMD-LSSVM-VMD-ELM carbon price prediction model based on error decomposition correction is proposed.The main steps are divided into three steps:(1)Data preprocessing;(2)Initial forecast;(3)Error decomposition correction prediction.Among them,the core of the model is to decompose the error sequence obtained from the initial prediction.This paper selects the price of carbon emission rights of China’s eight carbon trading pilots as a data set for an empirical analysis.In order to verify the prediction effect of the model,four sets of comparative experiment groups are set according to the model build step,and the model is verified by error evaluation index and the percentage indicator to verify the rationality and effectiveness of the structure,algorithm.By comparing the prediction result,the following conclusions are obtained:(1)Preprocessing data can reduce the interference of the data in data to predictive information,so it can significantly improve the single model prediction accuracy.(2)The prediction accuracy of the error decomposition correction model is significantly higher than that of other benchmark models.It is necessary to further decompose the error sequence and then use it to correct the initial prediction results.The error separation correction model integrates the advantages of decomposition method and error correction method,breaking through the prediction limitations of traditional combined models.Whether it is from prediction accuracy or stability,this model has obtained higher prediction accuracy in all carbon trading markets,and the prediction results are equipped with actual carbon prices,and the prediction level is further improved,which is carbon transaction.The system provides a practical carbon price prediction tool and research ideas,and has certain reference significance for the establishment and maturity development of my country’s carbon trading market.