Optimized LS-SVR method applied to vessel traffic flow prediction

This article firstly briefly introduces the principle of the non-linear Support Vector Regression Machine in the Support Vector Machines. Subsequently in order to improve and optimize the traditional Support Vector Regression Machine, Least Square Algorithm is adopted and Two-layer Planar Structure Optimization Method is put forward. Then the vessel traffic volume prediction model based on the optimized LS-SVR has been set up. Finally the prediction models are applied to the vessel traffic volume prediction of Yangtze Estuary Deepwater Channel, of which the volume is respectively obtained according to the characteristics of the length and gross tonnage of the ship. The performance comparison shows that the vessel traffic volume prediction model based on the optimized LS-SVR is valid and the model provides a good way for the medium-term prediction of traffic volume.