EU Carbon Price Predition Based On Multi-resolution Singular Value Decomposition And Double Optimization Algorithm

The carbon emission trading market is one of the effective measures to protect the environment and promote the low-carbon development of the economy.Its trading price is the core of the market mechanism.The accurate prediction of the price of EU carbon quota has become a research hotspot in the field of carbon emission in today’s world and provides valuable reference for market traders.This thesis presents a new hybrid optimization model MRSVD-PSO-SA –SVR,which has higher accuracy to predict the EU carbon quota’s price.First of all,the hybrid model adopts a new decomposition method,which is based on the principle of binary and matrix recursive generated,called multi-resolution singular value decomposition(MRSVD),then use particle swarm optimization(PSO)and simulated annealing(SA)intelligent optimization algorithm for support vector regression(SVR)model to optimize the parameters c and g;Lastly,select data from31 October 2010 to April 15,2019 daily closing price of the EU carbon quotas for empirical research.Through MSE,MAE,RMSE,MAPE,FVD,DC error analysis and the evaluation index,compared MRSVD-PSO-SA-SVR model with other models.The results show that the MRSVD decomposition method is superior to the singular spectrum decomposition(SSA)and wavelet decomposition(WD).Hybrid optimization algorithm of PSO-SA is superior to other optimization algorithms.Above all,The effect of the MRSVD-PSO-SA-SVR model proposed in this thesis has more accurate and stable prediction.