高速铁路非正常事件下初始延误场景聚类研究

Primary Delay Scenarios Clustering of High-speed Railway under Abnormal Events

在高速铁路日常行车组织工作中,及时准确地把握高速铁路非正常事件下的延误特征和事件分级是后续运行调整决策的基础。文章面对高速铁路非正常事件扰动,基于多源历史数据提取并分析初始延误场景参数特征,筛选初始延误时长和线路列车服务频率作为两个聚类指标,并应用轮廓系数论证FCM模糊聚类对当前场景的适用性。根据最终聚类结果,区间和车站非正常事件分别被聚为4类和3类。提出的聚类指标易于量化,非正常场景聚类分级结果能为实际延误管理工作提供有效支撑。

During the daily train organization of high-speed railway(HSR),grasping the delay characteristics and event classification timely and accurately is the basis of subsequent rescheduling decisions.Faced with the disturbance of HSR abnormal events,this paper first extracts and analyzes the parameter characteristics under primary delay scenarios based on multi-source historical data,then selects the primary delay and line service frequency as two clustering indicators,and validates the applicability of Fuzzy C-Means(FCM)clustering to current context via silhouette coefficient.According to the final clustering results,the abnormal events of section and station have been classified into 4 levels and 3 levels respectively.The proposed clustering indicators are easy to quantify,and the clustering results of abnormal events can provide effective support for the actual work of delay management.