Nowadays the interest in unmanned aerial vehicles (UAVs) is constantly increasing for different military and civil applications; these activities call for high-level requirements and therefore high-performance control laws. Mathematical models of the UAV dynamics can be used to synthesize the controllers. Classically, the problem is solved with a SISO approach, i.e., for each axis a model of the attitude dynamics is identified and then used to tune the parameters of a cascade control system. The aim of this thesis is to identify structured models for the angular and linear dynamics of a multi-rotor UAV, considering single-input-multiple-output (SIMO) problems. These structured models can be exploited to design control laws based on the UAV equations of motion. Black-box and grey-box model identification techniques are studied and applied to a quadrotor UAV. Then a new approach to obtain a structured model starting from an unstructured one has been proposed. Estimation data has been collected in flight, in closed-loop and in laboratory conditions at the Aerospace Systems and Control Laboratory of Politecnico di Milano. After analyzing and comparing the results, it was possible to design and simulate model inversion control laws such as the Explicit Model Following (EMF) and the Dynamic Inversion (DI). The designed controllers have been deployed to the quadrotor flight control unit and compared sending repeatable command inputs from a ground station. Experimental results have verified the achieved performance and have validated the accuracy of the identified models by comparing measured data with simulated data.
Al giorno d'oggi l'interesse per aeromobili a pilotaggio remoto (APR) è in continuo aumento in diverse applicazioni militari e civili, le quali prevedono requisiti sempre più stringenti e, pertanto, leggi di controllo ad alta prestazione. Per la sintesi dei controllori si possono usare modelli matematici rappresentanti la dinamica di questi velivoli. L'approccio comunemente utilizzato prevede l'identificazione di modelli della dinamica di assetto per il progetto di regolatori in cascata, considerando problemi di tipo singolo-ingresso-singola-uscita (SISO). Lo scopo di questa tesi è quello di ottenere modelli strutturati di tipo singolo-ingresso-multi-uscita (SIMO) della dinamica angolare e della dinamica lineare per un multirotore APR. Questo può permettere di progettare leggi di controllo basate sulle equazioni della dinamica del volo del velivolo. Tecniche di identificazione di modelli a scatola nera e a scatola grigia sono studiate e applicate a un quadricottero. Viene proposto inoltre un nuovo approccio per ottenere un modello strutturato partendo da uno non strutturato. I dati per la stima dei modelli sono stati raccolti in volo in anello chiuso presso il laboratorio di controllo e sistemi aerospaziali (ASCL) del Politecnico di Milano. Dopo aver analizzato e confrontato i risultati ottenuti, è stato possibile progettare e simulare le leggi di controllo quali la Dynamic Inversion (DI) e l'Explicit Model Following (EMF). I controllori progettati sono stati importati sull'unità di controllo di volo del quadricottero e confrontati comandando il drone da una stazione di terra. I risultati degli esperimenti hanno verificato le prestazioni raggiunte e validato l'affidabilità dei modelli identificati confrontando i dati misurati con quelli simulati.
Model identification and inversion-based control for multi-rotor UAVs
MIGLIORE, DANIELE
2018/2019
Abstract
Nowadays the interest in unmanned aerial vehicles (UAVs) is constantly increasing for different military and civil applications; these activities call for high-level requirements and therefore high-performance control laws. Mathematical models of the UAV dynamics can be used to synthesize the controllers. Classically, the problem is solved with a SISO approach, i.e., for each axis a model of the attitude dynamics is identified and then used to tune the parameters of a cascade control system. The aim of this thesis is to identify structured models for the angular and linear dynamics of a multi-rotor UAV, considering single-input-multiple-output (SIMO) problems. These structured models can be exploited to design control laws based on the UAV equations of motion. Black-box and grey-box model identification techniques are studied and applied to a quadrotor UAV. Then a new approach to obtain a structured model starting from an unstructured one has been proposed. Estimation data has been collected in flight, in closed-loop and in laboratory conditions at the Aerospace Systems and Control Laboratory of Politecnico di Milano. After analyzing and comparing the results, it was possible to design and simulate model inversion control laws such as the Explicit Model Following (EMF) and the Dynamic Inversion (DI). The designed controllers have been deployed to the quadrotor flight control unit and compared sending repeatable command inputs from a ground station. Experimental results have verified the achieved performance and have validated the accuracy of the identified models by comparing measured data with simulated data.| File | Dimensione | Formato | |
|---|---|---|---|
|
2019_10_Migliore.PDF.pdf
accessibile in internet per tutti
Descrizione: Testo della tesi
Dimensione
9.83 MB
Formato
Adobe PDF
|
9.83 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/149487