Research On Carbon Price Prediction Based On Secondary Decomposition And Deep Learning

Since the 21st century,the situation of global warming has become more and more serious.Reducing carbon dioxide emissions has become a hot topic in the international community.Facing the arduous task of carbon emission reduction,the international community promotes the reduction of carbon emissions by formulating climate policies and other measures.Among them,the carbon emissions trading mechanism has been proved to be able to effectively reduce carbon emissions and energy consumption.To achieve the goal of carbon emission reduction,China has steadily promoted the construction of the carbon emission trading market.Up to now,eight pilot carbon markets have been built in succession,and the actual trading of the national unified carbon market is about to open.The prediction of carbon emission trading price is an important part of carbon market trading.It has important theoretical and practical significance for the construction of the carbon market and the investment decision of market participants to research carbon price prediction,develop a better carbon price prediction model and improve the accuracy and robustness of carbon price prediction.To accurately predict the carbon price,this paper puts forward a hybrid model of carbon price prediction based on secondary decomposition algorithm and deep learning algorithm.This paper innovatively introduces the secondary decomposition algorithm into the field of carbon price prediction,which is the first time in this field.The main research contents of this paper are as follows.First,this paper summarizes the research status,development trends,and common carbon price forecasting methods at home and abroad,and points out the shortcomings of existing research.Second,the principles of empirical mode decomposition,variational mode decomposition,and long short-term memory network are introduced.Then,the carbon price prediction model is constructed.First,empirical mode decomposition is used to decompose the original data into several components,and then the high-frequency components are further processed by variational mode decomposition,to reduce the difficulty of prediction.Then all the components are predicted by the long short-term memory network,and the final prediction result is obt ained by summing the prediction results of each component.Then,through the empirical analysis of the actual data of all carbon market pilots in China,the proposed model is compared with other common models to verify the prediction accuracy and robustness of the model.Finally,the research results are summarized and analyzed,and the shortcomings and prospects are put forward.According to the research results,this paper summarizes four main conclusions.First,data preprocessing is necessary and effective to predict carbon prices with high nonlinearity and complexity.In this paper,through comparative experiments,it is confirmed that the direct prediction of historical data will lead to large errors.The prediction accuracy of the model combined with the data preprocessing method is generally better than that of the direct prediction model.Second,the secondary decomposition algorithm can effectively break through the limitations of the model and achieve higher accuracy prediction.The introduction of this method is feasible and reasonable,which expands the application of secondary decomposition in the prediction field.It has a certain reference value for improving the accuracy of carbon price prediction.Third,the proposed carbon price prediction m odel has the best prediction performance and robustness among all the comparison models.The experimental results further prove the superiority of the model in the field of carbon price prediction.It can be used as a powerful tool for carbon price predict ion.Fourth,the maturity of the carbon market in Hubei is the highest,while that in Tianjin is lower.