Design of robust nonlinear autopilot for helicopters with handling quality requirements

现在要求直升机具有高水平的敏捷性和机动性，以及在大范围空速的退化视觉条件下操作的能力。此外，出于认证目的，它们需要满足美国陆军航空设计标准ADS-33（更准确地说是ADS-33 E-PRF）规定的一些严格的可靠性和性能要求（所谓的操纵品质，HQ）。从控制工程师的角度来看，设计直升机飞行控制律是一个非常具有挑战性的问题：直升机呈现出高度非线性和完全耦合的动态特性，这些动态特性本身就是不稳定的。现有的数学模型从一个飞行区域到另一个飞行区域变化很大，并且由于气动力和力矩的经验表示而存在严重的不确定性。 尽管如此，对于今天飞行的大多数旋翼机，控制律已经主要使用经典的单回路频率响应和根轨迹设计技术来开发。这些技术是基于在飞行包线内的一组配平点的线性化模型，并且只提供局部稳定性。此外，为每个配平点改进飞行控制律，然后将它们与增益时间表"缝合"以覆盖整个飞行，这方面的费用和时间是相当大的，而且在飞行包线的极端部分实现一致的HQ还有很大的改进余地。本研究提出发展一种非线性自动驾驶仪，它不依赖于小信号近似，也不依赖于增益调度，是基于非线性自适应调节的最新成果。它能够在存在严重不确定性的情况下实现垂直/横向和纵向动态的精确跟踪。更重要的是，我们建议调查的手段，整合ADS-33为基础的要求，在自动驾驶仪的设计。该设计将在两个平台上进行验证：一个是我们实验室提供的类似于直升机的MIMO双旋翼系统，另一个是由魁北克公司Enciliity Inc.开发的直升机行业标准飞行模拟器HELISIM。

Helicopters are required nowadays to possess high levels of agility and maneuverability and the capability to operate in degraded visual conditions over a wide range of airspeeds. Furthermore, for certification purposes, they need to satisfy some stringent reliability and performance requirements (the so-called handling qualities, HQ's) set out by the US Army's Aeronautical Design Standard ADS-33 (more precisely ADS-33E-PRF). From the control engineer point of view, designing helicopter flight control laws is a very challenging problem: Helicopters present highly nonlinear and fully coupled dynamics that are inherently unstable. Existing mathematical models vary greatly from one flight region to the other and present severe uncertainties due to the empirical representation of aerodynamic forces and moments. Still, for most rotorcraft flying today, the control laws have been developed using predominantly classical single-loop frequency response and root locus design techniques. These techniques are based on linearized models about a set of trim points within the flight envelope and provide only local stability. Moreover, the cost and time associated with refining the flight control law for each trim point and then «stitching them together» with gain schedules to cover the full flight is substantial and there is still much room for improvement in achieving consistent HQ's in the extreme portions of the flight envelope. This research proposes to develop a nonlinear autopilot which does not rely on small signal approximation nor on gain scheduling and is based on recent results on nonlinear adaptive regulation. It is capable of achieving accurate tracking for vertical/lateral and longitudinal dynamics in the presence of severe uncertainties. More importantly, we propose to investigate means to integrate ADS-33-based requirements in the autopilot design. The design will be validated on two platforms: a MIMO Twin Rotor system resembling a helicopter available in our laboratory and HELISIM, an industry standard flight simulator for helicopters developed by the Quebec-based company Engenuity Inc.