INVERSE FLIGHT SIMULATION

Abstract

Flight simulation is a process of finding the aircraft outputs given the flight control commands. The inverse flight simulation problem, as its name implies is the inverse process of flight simulation. That is, from a given aircraft manoeuvre, determine the required flight control commands. In this thesis, the inverse flight simulation problem is studied. Since aircraft systems are generally non-minimum phase, both minimum and non-minimum phase systems are considered. The integration inverse simulation algorithm is a computationally efficient solution to the inverse flight simulation problem, and is capable of tracking the given trajectory closely. Unfortunately, this approach may generate unstable flight commands. This instability was attributed to the use of an unsuitable discretisation interval. The present research formulates a new stability test for the integration algorithm that enables us to predict the discretisation interval needed to stabilise the approach. A second disadvantage of using the integration algorithm is that it generates flight commands that can exceed their physical limits, i.e. unrealistic commands; and this may cause problem in the physical implementation. This is resolved using a new iterative local optimisation algorithm in which the flight commands are restricted within their physical limits with appropriate optimisation parameters. A stability test for the approach is also derived, thus allowing us to determine the discretisation interval needed for the flight commands to be stable. Compared with the integration algorithm, the local optimisation algorithm gives us less oscillatory flight commands; however, it is computationally less efficient. In practice, by applying the latter only when the former fails to generate realistic flight commands, we obtain a realistic and possibly computationally efficient solution. unrealistic commands; and this may cause problem in the physical implementation. This is resolved using a new iterative local optimisation algorithm in which the flight commands are restricted within their physical limits with appropriate optimisation parameters. A stability test for the approach is also derived, thus allowing us to determine the discretisation interval needed for the flight commands to be stable. Compared with the integration algorithm, the local optimisation algorithm gives us less oscillatory flight commands; however, it is computationally less efficient. In practice, by applying the latter only when the former fails to generate realistic flight commands, we obtain a realistic and possibly computationally efficient solution.