Guilhem Puyou

Sensitivity of the Generic Transport Model upset dynamics to time delay

Bifurcation analysis has previously been applied to the NASA Generic Transport Model (GTM) to provide insight into open-loop upset dynamics and also the impact on and sensitivity of such behaviour to closing the loop with a flight controller. However, these studies have not considered time delay in the system: this arises in all feedback controllers and has specific relevance when remotely piloting a vehicle such as the NASA AirSTAR GTM with ground-based controllers. Developments in the AirSTAR programme, in which a sub-scale generic airliner model will be tested for loss-of-control conditions over long ranges, raise the prospect of increased adverse effects of time delay relative to previous testing. This paper utilises bifurcation analysis, supplemented with time histories, on the GTM numerical model with a LQR-PI controller to evaluate the sensitivity of the closed-loop system stability to time delay. In this paper, the impact of time delays in both a fixed-gain and a gain-scheduled version of the controller is presented in terms of stability of nominal and off-nominal solutions.

Bifurcation Analysis of the NASA GTM with a View to Upset Recovery

The loss of control in-flight of civil airliners is a matter of great concern to the aviation industry. Loss of control in-flight often involves so called ‘upset’ conditions and, hence, the ability to recover from upset will reduce the frequency of loss of control incidents. This paper presents the use of bifurcation analysis, complemented by time-history simulations, to understand the flight dynamics of the open loop NASA Generic Transport Model with a view to identifying the attractors of the dynamical system that underlie upset behaviour. A number of drivers for potential upset conditions have been discovered which include non-oscillatory spirals and oscillatory spins. Time-histories of the upset conditions yield the response characteristics associated with these upset scenarios.

Upset Dynamics of an Airliner Model: A Nonlinear Bifurcation Analysis

Despite the significant improvement in safety linked to the fourth generation of airliners, the risk of encountering upset conditions remains an important consideration. Upset, which may arise from faults, external events, or inappropriate pilot inputs, can induce a loss-of-control incident if the pilot does not respond in the correct manner. Any initiative aimed at preventing such events requires an understanding of the fundamental aircraft behavior. This paper presents the use of bifurcation analysis, complemented by time-history simulations, to understand the flight dynamics of the open-loop NASA generic transport model by identifying the attractors of the dynamical system that govern upset behavior. A number of drivers for potential upset conditions have been identified, including nonoscillatory spirals and oscillatory spins. The analysis shows that these spirals and spins are connected in two-parameter space and that, by an inappropriate pilot reaction to the spiral, it is possible to enter the oscil...

Flight Control Law Design for a Flexible Aircraft: Limits of Performance

Limits of performance are explored when designing a flight control law for a flexible aircraft and especially the optimization of the wind comfort criterion under an actuator activity and roll-off constraint. When using a linear controller whose order is free, the issue is either to check the feasibility of design specifications, or to compute the maximal achievable performance, or to study the tradeoff between the design specifications. To this aim the Convex Control Design Toolbox is developed, whose aim is the synthesis of multiple-input/multiple-output feedback and feedforward controllers with H ∞ and H 2 specifications on a single plant model. Convex multimodel design is possible in the feedforward case, but parametric robustness specifications render the problem nonconvex in the feedback one.

Tolerance of aircraft longitudinal control to the loss of scheduling information : toward a performance oriented approach

Flight control laws are scheduled with respect to the flight point parameters. However systems failures can occur and induce a loss of these scheduling information. Today, it usually leads to switch to a lower level of automation that is more robust but less easy to handle (although aircraft is still safe). We would like to work on robust design methods to increase, as long as possible, the availability of functions that make flying task easy. This paper investigates the use of non-smooth optimization to solve a robust performance issue formulated as a minmax problem in a H∞ framework. Two algorithms have been benchmarked : HIFOO and HINFSTRUCT. From an application point of view, both algorithms provide real improvement of the robust performance; compared to the baseline controller designed using a worst case approach. Nevertheless HINFSTRUCT seems to be more reliable from a numerical point of view.

A multiobjective method for flight control law design

Due to developments in modern day aircrafts, requirements for flight control laws are increasingly numerous and varied. For this reason, control laws designers must now use multiobjective synthesis techniques. Our paper intends to develop such a method integrating both time and frequency domain constraints. We have chosen to develop a convex synthesis method to solve today’s aircraft issue. The whole procedure is based on the use of: first H1 control to compute an initial stabilizing controller, then Q-parametrization to introduce the closed loop shaping, convex synthesis to optimize the controller with respect to the requirements set and, finally, robust modal control to produce ecient and reduced order controllers.

Nonlinear Dynamics of Aircraft Controller Characteristics Outside the Standard Flight Envelope

In this paper, the influence of the flight control system over the offnominal behavior of a remotely operated air vehicle is evaluated. Of particular interest is the departure/upset characteristics of the closed-loop system near and beyond stall. The study vehicle is the NASA Generic Transport Model, and both fixed-gain and gain-scheduled versions of a linear quadratic regulator controller with proportional and integral components are evaluated. Bifurcation analysis is used to characterize spiral and spin behavior of the aircraft in closed-loop form and yields an understanding of the underlying vehicle dynamics outside the standard flight envelope. The use of a “gain parameter” to scale the controller gains provides information on the sensitivity of stability to gain variation, along with tracking how the controller modifies the open-loop steady states. Hence, this provides a means of assessing the effectiveness of the controller and evaluating the upset tendencies of the aircraft.

Feasibility of H∞ design specifications: an interpolation method

Given a set of H ∞ design specifications, the issue is to check whether there exists a controller, whose order is free, which satisfies these specifications. The classical solution, which is based on Youla parametrization and convex closed loop design, is not really satisfactory since it should use an infinite dimensional basis of filters, which cannot be done in practice. Let J* the minimal value of the design objective over such an infinite dimensional basis of filters. A Nevanlinna Pick interpolation method is proposed here to compute lower and upper bounds of J*, by solving the design problem on a finite set of frequencies. Finite-time convergence of the algorithm is proved.

Clearance Benchmark for a Civil Aircraft

This chapter describes the benchmark problem which will be used in the following chapters to address the clearance of flight control laws. Two sub-problems are considered. The first one, called ”nonlinear benchmark”, aims at validating the aircraft behavior close to the operating domain limits using a nonlinear rigid body model. The second one, called ”integral benchmark”, uses large scale linear models including flexible structural modes and represents a challenging problem for robust stability analysis methods. For each benchmark the underlying models and the associated clearance criteria to be validated are presented. The underlying models include detailed models of the flight mechanics or structural mechanics, together with the flight control laws to be assessed, as well as simplified models of the actuators and sensors. The clearance criteria cover a large range of certification requirements, as linear stability and performance analysis, or time domain performance evaluations. The current AIRBUS clearance methodology is also described and realistic expectations are stated regarding potential cost savings (both clearance effort and time) by using enhanced clearance technologies.

Gain-scheduled flight control law for flexible aircraft: A practical approach

Abstract This paper presents a gain-scheduling method applied to flight control law design. The method is a stability preserving interpolation technique of existing controllers under observer-state feedback form. Application is made on a flexible civil aircraft example considering multiple scheduling parameters. Although the interpolation technique gives powerful a priori stability guarantees, the sufficient condition to satisfy leads to conservative results in practice. We thus use a fixed observer model and check stability and performance thanks to μ-analysis. Provided results are really satisfactory for a final controller of little complexity.