基于管制通话语音个体特征的管制员不良工作状态识别方法研究

项目名称： 基于管制通话语音个体特征的管制员不良工作状态识别方法研究

项目编号： No.U1533117

项目类型： 联合基金项目

立项/批准年度： 2016

项目学科： 无线电电子学、电信技术

项目作者： 张兴俭

作者单位： 中国民航大学

项目金额： 34万元

中文摘要： 安全是民航运行的首要目标，空中交通管制员良好的管制质量是保障航空安全的前提。然而管制员可能出现的一些不良生心理状态直接降低其管制质量，造成安全隐患，而目前对此仍缺乏及时有效的应对措施。.课题充分考虑技术实用性、推广性和准确性，以可以方便获取的管制通话语音为基础，首次提出基于管制语音个体特征识别不良状态的研究思路，弥补了管制员状态识别研究不足。课题拟通过管制模拟实验获取典型不良状态（疲劳、高压或紧张）下的特征数据；开展管制语音受状态影响机理研究，各典型状态下管制语音变化特征研究以获取管制语音指标体系；并基于学习训练机制开展管制员个体语音特征库和个体特征建模方法研究；进而结合管制员语音变化特征和个体特征模型，建立基于管制语音个体特征的不良状态识别方法，并开发相应的管制员不良状态识别及预警系统样机。项目研究有望为减少管制员不良工作状态，提高航空安全水平提供支持，且具有广阔的理论与应用推广空间。

中文关键词： 空中交通管制员；人为因素；状态监测；特征提取；航空安全

英文摘要： Safety is the primary goal of the civil aviation. It is the premise of aviation safety for air traffic controllers to keep good control quality. However, some bad physical and mental states controller may occur directly impair his control quality, and this will cause potential safety hazard. Furthermore, there is no timely and effective response for this. .Based on the communication speech which can be collected expediently and considering the technical practicability, generalization and accuracy, the research approach “using controller’s individual speech to recognize bad states”, which can supplement the lack of the research on the recognition of controllers’ bad states, was proposed in this topic for the first time. Experiments will be conducted with the control simulation platform to collect data in typical bad states (fatigue, high pressure or nervous). The influence mechanism of bad states on control speech and the change characteristics of speech in each state will be studied to extract the indices of speech. Based on the training mechanism, the individual speech characteristics database will be established and the individual model method will be researched. Then, combining the change characteristics and the individual model, the recognition method of bad controlling state based on individual speech characteristic will be researched. Then, the prototypes of the controller’s bad states recognition and warning system will be developed based the above method. It is hoped that this study can provide a support for the decreasing bad control state and increasing aviation safety level. At the same time, there are wide space of promotion for the theory and the application of this research.

英文关键词： air traffic controller;human factor;state monitoring;feature extraction;aviation safety