基于虚拟人机系统的衰减记忆去伪控制器设计

Unfalsified Adaptive Controller Design with Fading Memory for Virtual Pilot-Aircraft System

对于包含人为因素飞行员模型的人机系统,传统针对确定模型的控制器方法无法完全满足飞行轨迹精度。针对飞行员模型在环系统,提出了采用衰减记忆方法、线性代价递增切换算法的去伪PI控制器设计方法。去伪控制理论是一种仅利用模型输入输出数据的无模型数据驱动控制理论。在去伪控制理论中,监视器用于选择最优控制器,控制器切换算法负责在需要时进行控制器切换。衰减记忆法的引入可减弱历史数据对切换判断的影响。对特定飞行场景进行了蒙特卡罗仿真,仿真结果表明,设计的控制器可减小飞行任务中的高度轨迹偏差,提高飞行品质。

For the pilot-aircraft system with pilot model concerning human factors, traditional control methods on specific models cannot totally satisfy flight trajectory precision. An unfalsified PI controller design method based on fading memory and linearly increasing cost level algorithm is introduced for such pilot-in-loop system. Unfalsified adaptive control theory is a non-model data-driven control theory, which only uses model input and output data. In the unfalsified control theory, the supervisor selects the best controller and the controller switching algorithm switches controllers when necessary. Fading memory method decreases the impact on switching algorithm which is caused by old data. Monte Carlo simulation is made in designed flight scenario. The result indicates that the designed controller improves the flight performance.