Sistem akustične detekcije brezpilotnih letalnikov

• Main Author: Papež, Matej
• Other Authors: Prezelj, Jurij
• Format: Master Thesis
• Language: Slovenian
• Published: 2017
• Subjects: brezpilotni letalnik psihoakustične cenilke lokalizacija vira hrupa klasifikacija hrupa uspešnost prepoznave meritve; drones psychoacoustic elevation of the source of noise sound classification sound recognition performance measurement; sami
• Online Access: https://repozitorij.uni-lj.si/IzpisGradiva.php?id=92958; https://repozitorij.uni-lj.si/Dokument.php?id=101125&dn=; https://repozitorij.uni-lj.si/Dokument.php?id=101126&dn=

Uporaba brezpilotnih letalnikov v zle namene stalno narašča. Razvili smo metodo, ki s pomočjo akustične analize signalov prepozna brezpilotni letalnik. Zasnovali smo merilno progo, ki je sposobna določiti elevacijo vira hrupa na osnovi uporabe mikrofonske antene, ter tudi obdelave psihoakustičnih značilk zvoka. Končna rešitev lokalizira in klasificira trenutno dominanten vir hrupa v okolici na podlagi primerjave vektorjev značilk v bazi znanih virov hrupa. Med izvajanjem eksperimenta smo preverjali na kakšni razdalji smo sposobni zaznati brezpilotnik, s kolikšno natančnostjo prepoznamo smer vira hrupa, ter kako uspešni smo pri sami prepoznavi vira hrupa. Vse meritve smo izvajali tako v gluhi komori, kot v realnem okolju. Ugotovili smo, da je detekcija brezpilotnika s prvim prototipnim sistemom lahko uspešna, vse pa je odvisno od količine hrupa okolice. Drones used for illegal purposes is a growing problem. The goal of the thesis is to create a drone detector, which can detect drones using audio analysis. We have designed a measuring line that is capable of determining the elevation of the source of noise based on the use of a microphone antenna, as well as the treatment of psychoacoustic sound characteristic. The final solution localizes and classifies the currently dominant source of noise in the surroundings based on a comparison of character vectors in the base of known noise sources. During the experiment, we checked on the distance from which we were able to detect a drone, how precisely we can recognize the direction of the source of noise, and how successful we are in identifying the source of a noise. All measurements were performed both in the deaf chamber and in the real environment. We found that the detection of a drone with the first prototype system can be successful, and everything depends on the amount of ambient noise.