Prediction of Ship Arrival Quantity Based on Optimized GM (1, 1) Power Model

Since the ship arrivals is vulnerable to climate, market season, economic development and other factors and is characterized for volatility, the prediction for ship arrivals by means of the conventional GM (1, 1) power model has such problems as low precision, non-relevance between the modeling method and the model testing. In this paper, the optimal matching model parameters for the ship arrival data are found and the GM (1, 1) power model with optimized model parameters is established using the nonlinear programming method, based on the conventional GM (1, 1) power model. Comparisons in term of accuracy in the application of the predictions are made between the ship arrivals data in Cen gang port area of Ningbo - Zhoushan port in May for the last 10 years and the predictions. The predicted ship arrivals for May, 2018 is 860, with an average relative error of 1.67%, and the average relative error of the optimized power model is greatly reduced, which shows that the average relative error of the optimized GM (1, 1) power model has been significantly reduced, with a relative residual not exceeding 7%. The results show that the optimized GM (1, 1) power model is able to further improve the prediction accuracy and meet the actual requirements, providing theoretical support for planning of the port anchorage.