Abstract: To accurately predict runway operational modes at multi-runway airports and enhance collaboration efficiency between airports and air traffic control departments, a machine learning method based on an ensemble model was proposed to predict runway operation modes. Firstly, the relevant concepts of runway operation mode were systematically sorted out, and the operation mode and its usage frequency of multi-runway airports were analyzed from multiple dimensions. Subsequently, three major categories of factors affecting runway operation modes and their ten sub-features were identified, and a runway operation mode prediction model was constructed by use of Voting ensemble classifier. Finally, empirical validation was conducted by selecting actud operational data from Beijing Daxing International Airport, covering September 2023 and March 2024. Experimental results demonstrate that the proposed ensemble learning model significantly outperforms traditional prediction methods in predication accuracy, particularly achieving breakthrough performance in the recognition accuracy of high-frequency operational modes, which attains prediction accuracy of 90.63%, 87.65%, and 83.56% in 1-hour, 3-hour, and 6-hour prediction task in advance respectively. These findings provide theoretical foundations for dynamic scheduling decisions while enhancing collaborative decision-making efficiency between airports and ATC entities.

Key words: traffic and transportation engineering; runway configuration; ensemble classifier; multi-runway airport

摘要： 为准确预测多跑道机场的跑道运行模式，提高机场与空管部门之间的协作效率，提出了一种基于集成模型的机器学习方法对跑道运行模式进行预测。首先，系统梳理了跑道运行模式的相关概念，从多维度分析了多跑道机场运行模式及其使用频率；然后，识别影响跑道运行模式的3大类因素及其10个子特征，利用Voting集成分类器构建了跑道运行模式预测模型；最后，以大兴机场为实证对象，选取了2023年9月及2024年3月的实际运行数据进行算例分析。实验结果表明，所构建的集成学习模型较传统预测方法在预测精度上实现显著提升，特别是在高频运行模式的识别准确率方面取得突破性进展，在提前1、3、6 h的预测任务中，预测准确率分别达到了90.63%、87.65%、 83.56%，可以为机场与空管部门之间的动态调度决策提供理论支撑，并提高协同决策效率。

关键词: 交通运输工程；跑道运行模式；集成分类器；多跑道机场