Combined Forecasting Model for Yangtze River Bulk Freight Index Based on VMD
Yangtze River Bulk Freight Index(YBFI)possess non-linear,non-stationary fluctuating traits which makes it difficult to predict accurately with traditional single prediction models and combination forecasting method.Therefore,following the idea of"decompose-restruct-subsequence forecasting-ensemble",a YBFI combined prediction model construction meth-od based on variational mode decomposition algorithm(VMD)was proposed.This paper decomposed the times series YBFI into high multiple modal components by using VMD.And then the modal components were reconstructed into high frequency,medium frequency,low frequency and trend sequences with clustering analysis method,and series fluctuating features were interpreted ac-cording to reconstruction outcome.Based on the comparison of multiple forecasting models,the high-frequency and low-fre-quency sequences are predicted using BPNN,and the medium-frequency and trend terms are predicted using the PSO-SVM method.Finally,integrated prediction value could be obtained by adding the reconstructed sequences predictions together.The empirical results showed that the combined forecast model based on VMD constructed in this paper has better prediction effect than single model,such as BPNN,SVM,PSO-SVM,ARIMA,PLS,and unoptimized VMD combined model as well as VMD-BP combination model.
万方数据
维普
