基于神经元簇功能假设的飞行员操纵行为习惯研究

The pilots' experience and the control behavior play an important role in fight safety. A proper pilot model to reflex the thinking process authentically, and a proper way to observe and analyse the pilot behavior more effectively will be of a great importance in aircraft design, fight training, rating and flight safety. In this items a kind of neural network cognition pilot model is developed based on the assumption of neural cluster. In which the human characteristics of online learning, experience system construction and memory mechanism are revealed by model simulation. As the hiding control habit, pilots' experience and other skilled factors can't be checked in the flight assessment criteria, a new method of testing is developed, in which both control ability, flight skill and mental workload are synthesized. In this valuation system the pilot's training level can be decided more accurately. Finally, the pilot behavior in landing and approaching are analysed by the pilot model, and a kind of effective training method to change the outmoded customs and construct the experience system is discussed. The research of pilot control habit in the item will play a significance role to promote the flight safety, raise the training efficiency and quality, in addition, this work has a potential value to reveal the mechanism of human learning, also the results of the item can be engaged in the artificial intelligence research.

飞行员的驾驶经验和操纵习惯对飞行安全有着重要影响。更加准确的建立飞机驾驶员行为模型、更加真实的反映人类思维过程、更加有效的对操纵行为进行监控与分析，对人机系统设计、飞行训练考核以及飞行安全有着重要意义。项目拟根据神经元簇功能假设，建立带有遗忘机制的飞行员认知控制神经网络模型，反映人类的在线学习、经验体系构建和记忆遗忘等特点，通过模型仿真进一步揭示人类学习行为和经验形成机理。同时针对现有飞行考核大纲无法对行为习惯和驾驶经验等非技能因素进行考核的问题，项目拟综合考量脑力负荷和飞行技术指标，建立行为习惯和驾驶经验评判标准，更加准确的对飞行员训练水平做出评判；最后项目将运用数学模型，对飞行员操纵行为进行分析，探索快速改变行为习惯、获取驾驶经验的高效训练方法。这一研究对促进飞行安全、提高训练水平具有直接推动作用，对揭示人类学习一般规律，开展人工智能研究具有潜在应用价值。

舰载机着舰下滑过程中，飞行员要不断观察引导光束、舱内仪表，同时反操纵驾驶杆和油门杆，在着舰指挥员指挥下，将飞机降落在网球场大小且不断运动的航母甲板上。这一过程中大气扰动频繁、飞行员工作负荷高、注意力分配难度大、突发情况多，尤其是下滑段飞机处于“反区”,着舰操纵方式与常规着陆存在显著差异，极易发生飞行事故。飞行员也普遍反映着舰驾驶行为习惯问题是训练中难点。因此开展飞行员驾驶行为研究，探寻飞行员驾驶习惯形成机理，分析飞行技能形成的一般规律，对促进舰载机飞行安全，开展飞行员高效训练有着重要意义。项目根据神经元簇功能假设，建立了带有遗忘机制的飞行员认知控制神经网络模型、飞行员自适应最优控制模型、飞行员注意力分配模型、着舰操纵策略模型、着舰飞行技能模型。结合飞行员的生理特点，相对全面的反映了舰载机飞行员耦合操纵、推理判断、操纵策略、注意力分配等特点。开展了飞行试验和数据采集，对舰载机飞行员着舰操纵技能成长情况进行了初步分析，结果表明不同经验分组的技能发展存在一定差异，同一经验分组的技能成长也存在着个体差异。总体来看，有三分之二的试验对象技能处于上升之中，三分之一的处于下降之中，在着舰技能成长方面，经验少的分组技能成长斜率通常大于经验多的分组，总体来看着舰飞行技能成长曲线形式上呈现幂指数下降形式。