基于K-means的铁路行车宏观安全风险动态预警区间确定

王列妮^1，2，胡士克²，黄毅²，曾琳惠²

1.西南交通大学交通运输与物流学院，四川 成都  611756；2.成都天佑安全技术有限公司，四川 成都  610031

摘要：为精确划分铁路行车宏观安全风险预警区间，避免仅依赖相关标准规范或人的经验确定安全风险预警区间，且预警区间不具备动态性的情况，建立宏观安全风险预警指标体系，采用K-means聚类算法确定宏观安全风险动态预警区间。通过K-means聚类算法将各指标集划分为好、中、差3类，以3类的上边界分别作为4个预警区间的分割点，确定各指标的4个风险预警等级的预警区间。通过实例验证该算法的可行性和适用性。结果表明：K-means聚类算法可确定铁路行车安全风险预警区间，并可根据历史数据变化动态调整。

关键词：铁路行车；宏观安全风险；K-means；预警区间

Determination of dynamic warning intervals for macroscopic safety risks in railway operations based onK-means

WANG Lieni^1,2, HU Shike², HUANG Yi², ZENG Linhui²

1. School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 611756, China；

2.Chengdu Tianyou Safety Technology Co., Ltd., Chengdu 610031,China

Abstract:To accurately divide the macroscopic safety risk warning intervals for railway operations and avoid relying solely on relevant standards or human experience to determine the safety risk warning intervals, as well as the lack of dynamism in the warning intervals, an indicator system for macroscopic safety risk warning is established. The K-means clustering algorithm is used to determine the dynamic warning intervals for macroscopic safety risks. The K-means clustering algorithm is employed to categorize different sets of indicators into three classes:good, moderate, and poor. The upper bounds of the three classes are used as the dividing points for the four warning intervals, to determine the warning intervals for the four risk warning levels of each indicator. The feasibility and applicability of this algorithm are verified through practical examples. The results show that the K-means clustering algorithm can determine the safety risk warning intervals for railway operations and can dynamically adjust them based on historical data changes.

Keywords:railway operation；macroscopic safety risk；K-means；warning interval