Interval Prediction Of Container Throughput Based On A Novel Hybrid Model EWT-PSO-SA-SVR

In recent years,with the continuous prosperity of Chinese economy,Chinese ports have become an important part of the world’s port transportation system.The accurate predicition of container throughput is very important for the construction and upgrading of ports.This paper focuses on the analysis of container throughput data of Dalian Port and Shenzhen Port,and proposes the hybrid model EWT-PSO-SA-SVR to perform interval forecasting.Firstly,the original data sequence is decomposed by an Empirical Wavelet Transform(EWT)algorithm to obtain a set of decomposed sub-sequences.Secondly,the Particle Swarm-Simulated Annealing Optimization algorithm was employed to optimize the parameters of Support Vector Regression(SVR).Finally,this paper use the proposed model EWT-PSO-SA-SVR to predict the container throughput of Dalian Port and Shenzhen Port.In this paper,the proposed model is also compared with other models,including EWT-CPSO-SVR,EWT-CS-SVR,EWT-PSO-SVR,and EWT-FA-SVR.The results show that the proposed hybird model has better performance than other models in predicting for container throughput data of Dalian Port and Shenzhen Port.