基于声学特征的管制员疲劳检测方法研究

Controller Fatigue Detecting Study Method Based on Acoustic Features

针对管制员工作疲劳问题,提出一种基于陆空通话分析的方法。利用雷达模拟机设计实验环境,采集一组包含PVT值和陆空通话记录数据。基于声学特征,分析了其对疲劳水平的贡献。随着疲劳程度的加重,各个参数均发生变化。通过SVM的基础上,验证了特征组合能够较准确地检测疲劳等级,为进一步形成管制员疲劳监测工具提供支持。

In this paper, a novel method based on radiotelephony records was proposed. Designing expe-rimental environment with ATC simulator, a bundle of data containing PVT values, and radi-otelephony records was collected. Based on selective acoustic features, analyzing of each feature in contribution to fatigue level has been done. It can be found that with the fatigue levels getting severe, changes exhibit in each parameter. Upon using SVM, it has also been proved that the combination of features is able to accurately detect fatigue level. The contributions of this paper could be used to construct applied tools for real-time controllers’ fatigue after further studies.