基于k-SC聚类的飞行操作模式及危险性分析

为挖掘飞行数据中的高危险性操作模式，采用k-SC时间序列聚类算法，并基于K-W检验分析聚类结果与不安全事件之间的关系。以特定条件下某机队着陆阶段的驾驶杆操作数据和长着陆事件为例进行模型验证。结果表明:机队有5类不同的驾驶杆操作模式；不同操作模式的着陆平飘距离分布具有显著性差异，且长着陆的危险性最高可达68.18%。研究结果可为提升航空公司飞行员操作能力、分析飞行员操作特性及其他不安全事件（如重着陆、擦机尾等）提供理论参考。

Analysis of flight operation patterns and risk based on k-SC clustering

In order to mine the high risk operation patterns of flight data,the k-SC time series clustering algorithm was used to analyze the relationship between clustering results and unsafe events based on the K-W test.The model was verified by taking the operation data of control column and long landing events during the landing stage of a certain fleet under specific conditions as an example.The results showed that the selected fleet had five different operation patterns of control column.There were significant differences in the landing drift distance distribution under different operation patterns,and the risk of long landing could reach up to 68.18%.The model provides theoretical support for further research on improving the operating capabilities of airline pilots,analyzing operating characteristics of pilots,and other unsafe events such as heavy landing and tail wiping.