Robust adaptive control: from theory to practical applications to aerospace systems

拟议的研究计划的总体目标是开发一种新的方法，用于设计具有自适应能力的鲁棒计划飞行控制系统，以及参加验证程序。 电子飞行控制系统的开发、集成和批准是一个繁琐、昂贵和耗时的过程。除了与稳定性、准确性和瞬态行为相关的标准要求外，EFCS还必须满足由飞行和操纵品质组成的严格规范，这些规范反映了飞机的易飞性。此外，EFCS必须对外部干扰（例如风、阵风）、变化的参数（例如质量、惯性、重心位置）以及潜在的数学模型不确定性和部分故障具有鲁棒性。 飞行器的动态特性在其飞行范围内变化很大，单一的线性时不变控制器不能保证在整个飞行包线内的鲁棒性能。几十年来，工程师们一直采用增益调度技术来设计飞行控制器。传统上，增益调度包括在飞行包线的网格上设计线性控制器;因此，控制器增益根据代表飞行条件（例如高度、马赫数、空速）的调度变量进行更新。控制器增益存储在查找表中，并使用多线性插值来计算每个操作点处的增益。这种方法提供了工作解决方案，但它是一个繁琐的迭代过程，涉及许多试验和错误循环。此外，所设计的控制器很少是最优的或鲁棒的。 该研究计划的主要目标是开发基于新理论方法的实用工具，以促进和改进航空航天系统的控制器设计。我们将重点应对以下三大挑战： 1.第一个挑战在于合成固定结构的控制器，可以很容易地调度，以覆盖整个操作域，以及控制方法，可以解决这个问题作为一个整体。为此，我们的第一个目标是利用迄今为止尚未开发的强大的理论成果，如监护人地图的鲁棒控制开发程序，产生强大的预定控制器的操作域。 2.第二个挑战涉及的全局稳定性评估的计划控制器。实际上，由于控制器合成通常在有限数量的操作点上执行，因此不能保证在每个其他点处的稳定性。因此，第二个目标提出了开发和使用新的工具的基础上监护人地图，以评估整个操作域的性能和鲁棒性。 3.最后，第三个挑战是在遇到不确定性或系统故障时恢复性能。对航空航天系统自适应方法的重新关注提供了新的工具，可以有利地用于增强基线控制器。为此，第三个目标的目的是增强鲁棒基线调度控制器与自适应回路。 这些新开发的方法和工具将大大有利于航空航天工业，降低开发成本，节省飞行控制系统的开发时间。

The overall goal of the proposed research program is to develop a novel methodology for the design of robust scheduled flight control systems with adaptation capabilities, as well as the attending validation procedures. The development, integration and clearance of electrical flight control systems (EFCS) are tedious, costly and time­-consuming processes. In addition to standard requirements related to stability, accuracy and transient behaviour, the EFCS must meet stringent specifications composed of flying and handling qualities that reflect the ease to fly the plane. Moreover the EFCS have to be robust to external disturbances (e.g. wind, gust), varying parameters (e.g. mass, inertia, centre of gravity location), and to underlying mathematical model uncertainties and partial failures. The dynamic behaviour of an aircraft varies greatly within its flight domain and one single linear time­-invariant controller cannot ensure robust performances over the whole flight envelope. For several decades, engineers have resorted to gain­-scheduling techniques to design flight controllers. Traditionally, gain-scheduling consists in designing linear controllers over a gridding of the flight envelope; controller gains are thus updated according to scheduling variables that are representative of the flight condition (e.g. altitude, Mach number, airspeed). The controller gains are stored in look-­up tables and multilinear interpolation is used to compute the gains at every operating point. This approach provides working solutions but is a cumbersome iterative process that involves many trials and errors loops. Moreover, the designed controllers are seldom optimal or robust. The main objective of this research program is to develop solid practical tools based on novel theoretical approaches that will facilitate and improve controller design for aerospace systems. We will focus on the three following major challenges: 1. The first challenge lies in the synthesis of fixed-structured controllers that can be easily scheduled to cover the entire operating domain, as well as control methods that can solve this problem as a whole. To this end, our first objective is to capitalize on hitherto unexploited powerful theoretical results in robust control such as the guardian maps to develop procedures that yield robust scheduled controllers with respect to the operating domain. 2. The second challenge concerns the global stability assessment of the scheduled controller. Indeed, as the controller synthesis is usually performed on a finite number of operating points, the stability at every other point is not guaranteed. Therefore, the second objective proposes to develop and to use novel tools based upon guardian maps to assess performances and robustness over the whole operating domain. 3. Finally, the third challenge consists in recovering performance versus encountered uncertainties or system failures. The renewed interest in adaptive methods for aerospace systems provides new tools that can be advantageously used to enhance baseline controllers. To this end, the third objective aims at augmenting the robust baseline scheduled controller with an adaptive loop. These newly developed methods and subsequently tools will greatly benefit the aerospace industry by reducing development costs and saving development time of flight control systems.