# Knowledge base: Warsaw University of Technology

Back

## Aircraft System Identification with simultaneous flight controls deflections

### Piotr Lichota

#### Abstract

This work describes a new method for multi-step inputs design in aircraft system identification. In the proposed manoeuvre the elevator, the ailerons and the rudder are deflected simultaneously as opposed to the typical approach. Identification was done with Quad-M methodology, that consists of four steps: Manoeuvre execution, signals Measurement, Model build-up and selection of the identification Method. The simultaneous design is based on the D-optimality criterion that requires a-priori knowledge about the system under test. This was found by an output error method linear system identification in the time domain. For the a priori model (initial iformation about the system) estimation typical system identification manoeuvres were used. Switching points in the multi-step D-optimal inputs were determined on the basis of the discrete wavelet transform for the a priori model outputs. The decomposed signals corresponding to each control surface deflection were weighted in order to determine points in which signals can change their state. The design variables were values of the input between switching points. The optimization task was done by a genetic algorithm with a linear ranking selection and elitism strategy. For the recombination the uniform crossover operator was chosen. In order to introduce more diversity in the population mutation was introduced as well. The D-optimal design was used as a set of input signals for the model and then a nonlinear system identification was performed. Nonlinear model parameters estimation was also done for the manoeuvres with harmonic multisine inputs and typical multi-step signals when flight control surfaces were deflected simultaneously or once at time. The multisine inputs were decorrelated in the time and frequency domains and optimized by the relative peak factor minimization what maximizes the input efficiency. The design with typical multi-step inputs applied simultaneously was optimized by the D-optimality criterion for the a priori model. Multisine manoeuvre and both designs with classical multi-step inputs had the same time-restrictions as the D-optimal flight surfaces deflections. In the end, D-optimal, multisine and both designs with typical inputs were compared in terms of parameters accuracy, manoeuvre time and acceleration. It was found that the D-optimal and multisine signals are superior to the designs with typical inputs and that they can be used in aircraft system identification. The evaluations were performed over Basic AirCraft Model in the Matlab environment.
Record ID
WUT9ac163d4179249b093c3b44b26b366b7
Diploma type
Doctor of Philosophy
Author
Title in Polish
Identyfikacja parametryczna modelu matematycznego samolotu z wykorzystaniem jednoczesnych wychyleń powierzchni sterowych
Title in English
Aircraft System Identification with simultaneous flight controls deflections
Language
(pl) Polish
Certifying Unit
Faculty of Power and Aeronautical Engineering (FPAE)
Discipline
mechanics / (technology domain) / (technological sciences)
Status
Finished
Start date
29-04-2014
Defense Date
29-09-2014
Title date
30-09-2014
Supervisor
Honored
yes
Pages
133
Keywords in English
Flight Dynamics, System Identification, Input Design, Genetic Algorithm, Wavelet Transform
Abstract in English
This work describes a new method for multi-step inputs design in aircraft system identification. In the proposed manoeuvre the elevator, the ailerons and the rudder are deflected simultaneously as opposed to the typical approach. Identification was done with Quad-M methodology, that consists of four steps: Manoeuvre execution, signals Measurement, Model build-up and selection of the identification Method. The simultaneous design is based on the D-optimality criterion that requires a-priori knowledge about the system under test. This was found by an output error method linear system identification in the time domain. For the a priori model (initial iformation about the system) estimation typical system identification manoeuvres were used. Switching points in the multi-step D-optimal inputs were determined on the basis of the discrete wavelet transform for the a priori model outputs. The decomposed signals corresponding to each control surface deflection were weighted in order to determine points in which signals can change their state. The design variables were values of the input between switching points. The optimization task was done by a genetic algorithm with a linear ranking selection and elitism strategy. For the recombination the uniform crossover operator was chosen. In order to introduce more diversity in the population mutation was introduced as well. The D-optimal design was used as a set of input signals for the model and then a nonlinear system identification was performed. Nonlinear model parameters estimation was also done for the manoeuvres with harmonic multisine inputs and typical multi-step signals when flight control surfaces were deflected simultaneously or once at time. The multisine inputs were decorrelated in the time and frequency domains and optimized by the relative peak factor minimization what maximizes the input efficiency. The design with typical multi-step inputs applied simultaneously was optimized by the D-optimality criterion for the a priori model. Multisine manoeuvre and both designs with classical multi-step inputs had the same time-restrictions as the D-optimal flight surfaces deflections. In the end, D-optimal, multisine and both designs with typical inputs were compared in terms of parameters accuracy, manoeuvre time and acceleration. It was found that the D-optimal and multisine signals are superior to the designs with typical inputs and that they can be used in aircraft system identification. The evaluations were performed over Basic AirCraft Model in the Matlab environment.
Thesis file
• File: 1
Praca doktorska.docx
Request a WCAG compliant version

Uniform Resource Identifier
https://repo.pw.edu.pl/info/phd/WUT9ac163d4179249b093c3b44b26b366b7/
URN
urn:pw-repo:WUT9ac163d4179249b093c3b44b26b366b7

Confirmation
Are you sure?