Forecast Application of Time Series Model Based on BLS in Port Cargo Throughput

Abstract

During the last decade, there was a dramatic increasing of container throughput in China, especially in server harbor cities such as Shanghai and Shenzhen. It is a necessary and crucial task to enhance the ability of port throughput. In the existing studies, time-series model is one of the most powerful methods to solve this problem, which can predict the container cargo throughput accurately and effectively. Based on this technical background, this paper employs a novel algorithm to design a new type of time-series model for predicting port throughput. In the experiments, we firstly use the Matlab to pursue the statistical analyses on the throughput data. Secondly, we apply our method to predict the changing rate of container throughput, and compare the results with several classic time-series models. Finally, the experimental results show that our method was optimized based on the training data, and outperformed other time-series models in the prediction of 10 months throughput.