Study On The Price Prediction Of Carbon Emission Right Trading Based On CNN-LSTM Model

Carbon price is the core of the carbon financial market operation,with the development of carbon financial market,carbon market risk is concerned,in terms of carbon price scenario,carbon price level itself to improve environmental quality,reduce energy demand and improve macroeconomic growth of different influence,developed a reliable carbon price prediction technology to promote the development of Chinese carbon financial market mature is of great significance.In this paper,the price data set of Hubei carbon market is selected as a sample for carbon price prediction,and the LSTM is used as the benchmark model to further construct the CNN-LSTM model containing one-dimensional convolutional layer(Conv1D),a maximum pooling layer(Max Pooling1D),an LSTM layer and a fully connected layer(Dense).Finally,the two carbon price prediction models and the reference model are compared.Discovery:(1)the generalization ability of the LSTM model needs to be improved,CNN-LSTM model can effectively improve the prediction accuracy and training speed of carbon price in Hubei market;(2)If the important index characteristics of carbon price in Hubei pilot are selected as data set,Using the CNN-LSTM model for the prediction,The time step is the historical data length selected as 3,To achieve the optimal result of Hubei carbon price forecast,This is because the goodness-of-fit R2 based on this result reaches 0.9843,The prediction error of MAPE was only 0.24%,And the prediction time is only about 0.15 seconds,A total of five evaluation indicators all performed the best;(3)The CNN-LSTM model combines two different types of 1 dimensional convolution and LSTM,Ability to exploit the local feature extraction of the convolutional layer and the time-series modeling capabilities of the LSTM layer,Further improve the performance of the model,at the same time,Pooling layers can reduce the output size of the convolution layers.In addition,this paper selects the average price BEA of Beijing carbon emission right for case analysis,and verifies the superiority of CNN-LSTM model in the accuracy and applicability of carbon price prediction from the perspectives of single feature prediction and multi-feature prediction.