EMD Decomposition Integrated Combination Forecasting And Its Application In Carbon Trading Prices

“The development of sustainable and the protection of environmental”are the 21st century subjects.However,with the destruction of the ecological environment,the discharge of a large number of waste gases in daily production and life,and the rapid growth of the global population in recent years.These factors lead to global warming,which poses a very serious threat to human life.Climate warming is a worldwide problem,all countries in the world are making efforts for environmental protection,and our country regards environmental protection as a fundamental national policy to which our country has always adhered.According to statistics,air pollution is one of the most major environmental issue in China.The rapid development of industry and the dramatic increase of population are the reasons for the increasing air pollution in our country,the desertification of green land,the ravages of haze weather and the vulnerability of summer floods are all caused by environmental pollution.Among the causes of environmental pollution,the most important is carbon emissions,so controlling carbon emissions has become an important measure of environmental protection.Carbon trading is a kind of trading measure to control greenhouse gas emissions in the world.By monitoring and restricting carbon trading,greenhouse gas emissions can be effectively controlled,thus global warming can be suppressed and improved.China's current carbon trading markets include Beijing Municipality,the Province of Fujian,the Province of Guangdong,the Province of Hubei,Shanghai Municipality,Shenzhen Municipality,Tianjin Municipality and Chongqing Municipality.This article obtained the carbon trading price related data through the Python at the Chinese carbon trading website(http://k.tanjiaoyi.com).According to the data of 2020,we can see that Guangdong Province is the first of the eight carbon trading markets in China in terms of total carbon trading and total carbon trading.Therefore,this paper selects Guangdong Province 2020 carbon trading price to forecast and analyze.Because of the volatility,irregularity and nonstationarity of carbon price time series in complex dynamic markets,if the prediction method is directly used to predict it,the result is not very ideal,therefore in order to obtain more accurate prediction values,a"first decomposition and then integration"processing method is adopted.Firstly,the EMD empirical mode decomposition method is used to decompose the carbon trading price data of Guangdong Province,After decomposition,several IMF components and a Res components are formed,and then calculate the average value of each IMF component,according to whether the average value is near 0 to distinguish the high frequency components and low frequency components,In order to obtain the low frequency sequence and the trend term sequence,the components of the mean value in the 0 attachment are added recombined into high frequency sequence,the low frequency sequence is obtained by adding the components whose mean value is not near 0;next,four single prediction methods(exponential smoothing,ARIMA(differential Integration Mobile Average Autoregression),LSTM(long-term short-term neural network),and SVM(support vector machine)were used to predict the different frequency sequences,which has been decomposed and recombined,Four new methods can be obtained by summing the predicted values of the different frequency sequences;Then,based on theL_P norm as the optimal criterion,based on L_P norm carbon trading price,the combined weighted coefficient prediction model is constructed under two special values,On this basis,take into account the difference in the prediction effect of each single prediction method at each point in time.Therefore,the introduction of information collection and settlement is introduced into the model which was based on the fixed-weight coefficient combination forecasting model,On this basis,the sum of the absolute value of error,and the square root of the sum of the error squared are all introduced.The combination predictive model with variable weight coefficient which was based on he square root of the sum of the error squared of the IOWA operator and the combination prediction model with variable weight coefficient which was based on sum of absolute value of error of IOWGA operator are constructed;Finally,the combined prediction model of fixed weight coefficient,and variable weight coefficient which was based on information aggregation operator are applied to the carbon trading price data of Guangdong Province in 2020,As can be seen from the experimental results,The constructed fixed-weight coefficient combination forecasting model,and the variable-weight coefficient combination forecasting model which was based on the information collection and settlement sub-predictor have more better forecasting effects than the single forecasting method,and also have more better fitting effect,and the combination prediction model is excellent combination prediction.