飞行员疲劳度评估模型及应用研究

发布时间：2018-11-13 07:21

【摘要】：控制飞行人员的疲劳状态是飞行安全的重要课题,目前的飞行员疲劳状态研究更多停留在归纳总结、建议或者生物学模型之上,并没有做到真正的建模与量化,利用信息管理系统以及机器学习算法控制飞行人员的疲劳状态是本文的核心内容。本文结合软件工程的理论知识,根据航空公司对飞行疲劳监控的需求进行分析,针对需求分析进行了系统的总体设计,系统归纳为信息管理、疲劳值管理、航班推荐管理和航后休息期管理四个子模块,在详细设计阶段我们通过参阅公司的相关运行规范以及咨询飞行专家进行了归纳和总结,确定了飞行员不同主客观因素产生的疲劳值情况;在此基础上,将多种疲劳因素引入线性回归模型进行建模,得到疲劳值模型;通过构建飞行员-航班疲劳值矩阵,利用PMF模型进行航班推荐建模;结合民航总局规定的基本的航后休息期,将飞行员的疲劳值量化到航后休息期中,构建航后休息期优化模型;最终将三个模型整合到飞行员疲劳监控管理系统中来,形成统一的管理,以期全面量化的分析和监控飞行员的疲劳状态,控制疲劳值,使得飞行安全有保障。本文利用实际采集到的飞行人员的航班数据以及真实的航班运行条件,并通过专家评估的方式对模型进行了训练与论证,最终整合到整个系统应用中,并利用业界流行的Spring MVC框架及python算法库形成了一套基于B/S框架的飞行人员疲劳监控系统,并验证了其可行性与可用性。
[Abstract]:Controlling the fatigue state of the pilots is an important subject of flight safety. The current research on the fatigue status of pilots is more focused on summing up, suggesting or biological models, and does not achieve real modeling and quantification. Using information management system and machine learning algorithm to control the fatigue state of pilots is the core of this paper. Based on the theoretical knowledge of software engineering, this paper analyzes the requirements of flight fatigue monitoring and control by airlines, and makes the overall design of the system for demand analysis, which is summarized as information management and fatigue value management. The four sub-modules of flight recommendation management and post-flight rest period management have been summarized and summarized by referring to the company's relevant operating specifications and consulting flight experts in the detailed design stage. The fatigue values of pilots caused by different subjective and objective factors are determined. On this basis, a variety of fatigue factors are introduced into the linear regression model to model the fatigue value model, and the pilot-flight fatigue value matrix is constructed, and the PMF model is used to model the flight recommendation. Combined with the basic post-flight rest period stipulated by CAAC, the fatigue value of the pilot is quantified to the post-flight rest period, and the optimization model of post-flight rest period is constructed. Finally, the three models are integrated into the pilot fatigue monitoring and management system to form a unified management, in order to analyze and monitor the pilot's fatigue state and control the fatigue value in order to ensure the flight safety. This paper makes use of the flight data collected by the flight personnel and the real flight operation conditions, and through the way of expert evaluation, the model is trained and demonstrated, and finally integrated into the application of the whole system. Using the popular Spring MVC framework and python algorithm library, a flight fatigue monitoring system based on the B / S framework is developed, and the feasibility and availability of the system are verified.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：V328;TP311.52

【相似文献】