疲劳状态下陆空通话语音特征变化研究

为预防空中交通管制不安全事件,利用陆空通话语音数据结合精神运动能力检测(PVT)方法,研究管制员的疲劳预警机制。首先,通过处理管制模拟机训练过程中陆空通话的音频文件,截取有效的管制指令信息,提取重新定义的语速、音调等特征;然后,采用PVT方法测试管制员在每次训练前后的反应能力,对比各个语音特征和PVT测试结果;最后,采用箱型图统计观察各时间段语音特征的变化趋势,并通过k-means++算法分析疲劳状态下绝对和相对语音特征区间,观察管制员工作中疲劳的变化过程,提出安全红线。研究表明:所提出的方法可以通过语音特征有效监控管制员的工作状态,为人员疲劳进行预警,从而实现管制不安全事件的预防,提高安全管理水平。

Study on voice feature change of radiotelephony communication under fatigue state

In order to prevent air traffic control incidents,fatigue warning mechanism of controllers is studied by using voice data of radiotelephony communication and PVT method. Firstly,effective control instruction information was intercepted and redefined speed speed,tone and other features were extracted from audio files of radiotelephony communication during training process of control simulator. Then,controllers’ response capability before and after each training was tested by using PVT method,and speech features and PVT test results were compared. Finally,change trend of speech features in each time period was observed with box chart statistics,absolute and relative speech feature intervals under fatigue state were analyzed by k-means ++ algorithm,and safety red line was proposed based on observation on change process of controllers ’ fatigue. The research shows that the proposed method can provide early fatigue warning and therefore prevent incidents and improve safety management by effectively monitoring controllers’ work state through their voice features.