Electronic Theses and Dissertations

Date

2023

Document Type

Thesis

Degree Name

Master of Science

Department

Electrical & Computer Engineering

Committee Chair

Eddie Jacobs

Committee Member

Daniel V Foti

Committee Member

Alfredo J Ramirez

Abstract

Acoustic sensors are devices that are not commonly used on autonomous uncrewed aerial vehicles (UAV). Obtaining a usable signal-to-noise ratio (SNR) is challenging. Given the most problematic noise is the flight-induced wind noise, one way of approaching the problem is to stop the wind noise at the source by designing a mount for the acoustic sensors to reduce the wind component before the signal and noise enter the microphone. Subsequently, signal processing stages can be added to improve the SNR further. We begin by formulating an atmospheric attenuation model using both point and line acoustic sources. The model predicts the frequency spectrum and how it reacts to changes in atmospheric conditions. This model is used to predict the SNR over the frequency range of interest as measured at the UAV for various wind speeds for a given acoustic source sound pressure level (SPL) as well as predict the SNR as a function of distance. Multiple fixed-wing UAV mounting strategies are then developed based on the predicted airflow during flight with each analyzed with respect to SNR. Based on predicted SNRs, various signal processing algorithms are evaluated for their improvement of detection statistics. Finally, the SNR of the processed signal is evaluated for usability. Particular instantiations of the acoustic sensing wing mounts are evaluated in the lab using a wind tunnel as well as in some physical UAV test flights. Data collected from these flights is processed offline using different signal processing approaches. Based on the model predictions and the results of the limited field measurements, conclusions regarding the feasibility of acoustic sensing on a UAV are discussed.

Comments

Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to ProQuest

Notes

Open Access

Download

Share

COinS