飞机过站上客过程持续时间分布

Boarding duration distribution of aircraft turnaround

为了准确预估飞机过站上客过程持续时间,研究了上客过程持续时间的随机性与可捕获的规律性,结合北京首都国际机场运行情况,分析了影响上客过程持续时间主要因素,根据机型、停机位和到港时间对飞机过站数据进行分组,统计了上客过程持续时间的频数,分别采用正态分布、韦布尔分布和泊松分布对上客过程持续时间进行分布拟合,通过卡方拟合优度检验分布,采用最大似然估计法对分布的未知参数进行估计,得到上客过程在各个组合条件下的完整分布,构建了飞机过站上客过程持续时间分布模型。分析结果表明:3种不同分布对上客过程持续时间都有较好的拟合度,韦布尔分布适用于各种组合条件,但在分组Ⅲ(中型机、近机位、22:00~23:59)、Ⅴ(中型机、远机位、09:00~21:59)、Ⅶ(大型机、近机位、00:00~08:59)、Ⅷ(大型机、近机位、09:00~21:59)和Ⅸ(大型机、近机位、22:00~23:59)中,正态分布与泊松分布的拟合优度大于韦布尔分布的拟合优度;所建分布模型能很好地预测上客过程持续时间,上客过程持续时间实际值与根据分布模型计算的数值的平均误差小于2min;置信水平为90%的分布区间宽度为10~12min,平均预测准确率为83.63%,置信水平为75%的分布区间宽度为7~9min,平均预测准确率为66.94%,可以看出分布区间的预测准确率几乎达到了置信水平,因此,采用分布模型可准确预测飞机过站上客过程持续时间。

In order to accurately predict aircraft turnaround boarding duration, the randomness and catchable regularity of boarding duration were studied. Combining the operation situation of Beijing Capital International Airport, the aircraft turnaround data were grouped according to aircraft type, gate position and arrive time on the basis of analyzing the main factors influencing on the boarding duration. The frequency of aircraft boarding duration was counted, and the distribution of boarding duration was fitted by using normal distribution, Weibull distribution and Poisson distribution. Chi-square fitting goodness was used to test each distribution, the unknown parameters of the distributions were estimated by the maximum likelihood estimation method, the complete distributions of boarding duration in different combined conditions were got, and the distribution models of boarding duration were constructed. Analysis result shows that the distribution of boarding duration can be fitted by the three distributions, and the fitting goodnesses are larger. Weibull distribution adapts to every group of data. However, normal distribution and Poisson distribution have better fitting goodnesses in groups Ⅲ（medium-sized aircraft, contact stands, 22：00-23：59）, Ⅴ（medium-sized aircraft, remote stands, 09：00-21：59）, Ⅶ（large-sized aircraft, contact stands, 00：00-08：59）, Ⅷ（large-sized aircraft, contact stands, 09：00-21：59）, and Ⅸ（large-sized aircraft, contact stands, 22：00-23：59）. The distribution models can be used to forecast the boarding duration well, and the average error of the actual values and the calculated values is less than 2 min. When the confidence level is 90%, the distribution interval width is 10-12 min, and the average prediction accuracy is 83.63%. When the confidence level is 75%, the distribution interval width is 7-9 min, and the average prediction accuracy is 66.94%. It can be seen that the accuracy of the distribution interval almost reachs the confidence level. Obviously, the distribution models can be used to predict the aircraft boarding duration accurately. 5 tabs, 8 figs, 22 refs.