非齊性卜松過程分析貨櫃船之到達過程~以台中港為例

Abstract

港埠分析主要須先得知船舶到達率，若能精確的描述其到達過程，則更能掌握實際港埠營運情形。卜松過程的特性為任一時間點之船舶到達率為常數，此過程在描述一成長或衰退之模式時並無法表現其趨勢，故可以非齊性卜松過程描述之。非齊性卜松過程最主要特點為船舶之到達率為時間的函數，到達率亦即強度函數不僅與所橫跨之時段長有關，且與所在之時間點有關，非齊性卜松過程較不似齊性卜松過程具有僵固性，故此時使用非齊性卜松過程分析更能符合實際資料情形。 本研究以非齊性卜松過程分析貨櫃船到達過程，根據台中港貨櫃船到港資料進行分析，強度函數以最常用之冪定律強度函數及對數線性強度函數模式進行分析，以最大概似法求參數估計值，並進行參數檢定及模式適合度檢定。模式適合度檢定結果顯示冪定律強度函數與對數線性強度函數模式皆可描述台中港貨櫃船到達過程，兩者結果一致皆呈現台中港貨櫃船之緩慢成長趨勢。再依AIC法則選出對數線性強度函數模式為適合度較佳之模式。

It is necessary to know the arrival rate of ships before port analysis. If we can describe the arrival process of ships, we can control more the real situation of operating a port. The characteristic of a homogeneous Poisson Process (HPP) is the arrival rate of a ship at anytime is a constant and HPP can’t describe the time trend. Instead of using a HPP to describe a developing or decline model, we can use a non-Homogeneous Poisson Process (NHPP) to illustrate. The main characteristic of NHPP is its arrival rate is the function of time t. The arrival rate can be called as intensity function. It is not only related to time intervals but also related to time t. A NHPP is more fixable than a HPP. Therefore, if we use a NHPP to analyze, the result can more correspond to the real situation.

This research was aimed to analyze the arriving process of containerships by NHPP. The research resource was based on the data of Taichung Harbor. Intensity function is analyzed by the power law models and the log-linear models. The maximum likelihood parameter estimation and goodness-of-fit analysis are also applied. The results of goodness-of-fit analysis showed that both power law models and the log-linear models can describe the process of the containerships from Taichung Harbor. Both results indicated that the containerships in Taichung Harbor is slowly developing.It can be decided by AIC law that the log-linear model is more preferable in this case. As a conclusion, the NHPP is a suitable tool for analyzing the arrival process of containerships.