Model aircraft to improve flight testing

Roger Larsson presents a doctoral thesis in which he has looked at several methods to optimise the modelling methods used to test and develop aircraft. He has also used model airplanes from colleagues.

Roger Larsson Roger Larsson

In the aircraft industry, models and simulations are used to investigate the properties of aerial vehicles, to improve and refine their flight behaviour. Modelling saves both time and money, provided that the model selected gives accurate results. The science used to develop the underlying mathematical models is known as system identification. The researchers create models of dynamic systems with the aid of data that have been collected.

“Gripen, for example, is equipped with many sensors that provide us with data during a complete test flight, and we can use the data to improve existing models. It’s our goal to be able to use optimisation to develop better models to describe the motion of the aircraft in various conditions”, says Roger Larsson, who recently presented his doctoral thesis in the Division for Automatic Control.

Identification algorithms

He has looked at two types of identification algorithms: linear and non-linear. Linear algorithms are intended for use when the aircraft is in the air, and provide the engineer with information that can be used to take decisions during flight. If the data collected are consistent with the existing models, testing continues. If they do not agree, detailed analysis must be carried out. Roger Larsson has refined and improved an existing method to make it more robust and able to cope with different types of disturbance – such as turbulence or measurement errors.

The collaboration within the LINK-SIC research programme led to Roger making contact with colleagues Alejandro Sobron and David Lundström, in the Division of Fluid and Mechatronic Systems, and they hatched the idea of conducting test flights with their model airplanes. These tests were subsequently carried out together on the Bråvalla aerodrome. In particular, the work gave important insights into the choice of linear model.

Non-linear models, in contrast, concern identifying the more advanced properties of an aerial vehicle. The data are analysed when the test has been concluded and the analysis concerns phenomena that are more difficult to deal with, such as tight turns, loops and severe turbulence.

“And this was the research I worked with initially when I started my doctoral studies in 2007”, says Roger Larsson. During all eleven years of his industry-based doctoral work he has also worked at Saab Aeronautics.

Interesting collaborations

He sees mainly advantages in having been a doctoral student for such a long period.
“It’s given me space to think, and see how all the pieces fit together. On the other hand, of course, I’ve forgotten some of what I did at the beginning as time has passed. It’s good that I have my licentiate thesis to read when I need reminding”, he laughs.

“These have been productive years, and I’ve been involved with the LINK-SIC research collaboration right from the start. It’s interesting when the problem formulations come from industrial actors, and we receive research questions from other industries. I feel familiar with these topics, but at the same time gain new ideas to test. It was also interesting to come from the field of automatic control into the world of fluid dynamics. Aircraft development has undergone huge developments, and is currently poised at the interface between automatic control and fluid dynamics. When these two fields collaborate, we see that 1 + 1 can be as much as 3½”, says Roger Larsson.

It’s with something of a heavy heart that he is leaving his office at LiU, but other collaborative projects will arise in the future.

“Oh yes, I’ve got several ideas, and I hope that the theoretical part of my thesis can be used in other contexts. I’m happy when our work can be put to practical use.”

Flight Test System Identification, Roger Larsson, Division for Automatic Control, Department of Electrical Engineering, Linköping University 2019.
Principal supervisors: Dr Martin Enqvist and Professor Emeritus Lennart Ljung

Contact
Show/Hide content

Research
Show/Hide content

Latest news from LiU
Show/Hide content