Container throughput forecasting of Nansha Port based on ARIMA-RBF

Abstract

Predicting container throughput is fundamental for port management and the scheduling of handling equipment. Based on an analysis of the mechanisms of the ARIMA model and the RBF model, this paper investigates the daily container throughput patterns and data stationarity characteristics of Nansha Port in Guangzhou, using survey results. By integrating the time series forecasting capability of the ARIMA model with the nonlinear processing ability of the RBF neural network, an ARIMA-RBF combined forecasting model is established to predict the container throughput of Nansha Port. This model accounts for both the linear and nonlinear characteristics of port container throughput and demonstrates superior predictive performance compared to the traditional ARIMA forecasting model.