The Automated Flight Control System (AFCS) of a transport category aircraft is made up of several modes. Some are used to establish a desired attitude, which could be commanded directly by the pilot or through outer loops which are designed to track a trajectory. To achieve this goal for the entire flight envelope, aircraft manufacturers must generate a series of control laws which are only valid for a narrow operational range of speed and altitude. With the dynamic inversion approach, a single control law can satisfy handling qualities requirements for the complete flight envelope.
In this research project, the focus is put on four control axis which are normally used continuously by the flight guidance computer when the autopilot is engaged: pitch angle and roll angle command; auto-thrust; yaw damping and turn coordination. The objective is to validate whether a model estimation in a controller designed with dynamic inversion would meet the flying qualities requirements commonly used in the industry. This simplification of the control laws for transport category aircrafts would mean significant savings in cost and time required for the design of an autopilot.
A Boeing 747 model was created which enabled a high level of flexibility for online tuning of the control laws, reducing the complexity of troubleshooting. The controller was then implemented on an independent simulation platform called FlightSIM, to evaluate the reproducibility of the preliminary results. Test procedures were elaborated to allow measurements for a time analysis of the simulation data, throughout the operational flight envelope.
The global assessment showed a satisfying level of performance of the handling qualities, for the entire envelope. The longitudinal controller exhibited superior flying characteristics compared to the lateral controller. With the analysis of the results, it was concluded that the improvement of the performance was a consequence of the integration of the dynamics of actuators. With this added feature, the command law derivation yielded equations which would compute an angle correction for the control surface, rather than estimating the exact angle of deflection required to reach the reference signal. With this approach, the aircraft preselected dynamics correlated better with the observed time response, and the steady-state error was significantly reduced.
| Date | 11 Jun 2018 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Ouassima Akhrif (Supervisor) & Franck Cazaurang (Co-supervisor) |
|---|
Brisset, J. N. (Author),
Akhrif (Supervisor) & Cazaurang (Co-supervisor),
11 Jun 2018Student thesis: Master's thesis › Master in Engineering: Engineering