| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 2.29 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Atualmente, diversas operações subaquáticas realizadas em mar aberto ou rios são efetuadas por sistemas dedicados, recorrendo regularmente, ao auxílio de ROV’s (Remotely Operated Vehicles) e AUV’s (Autonomous Underwater Vehicle). No desenrolar destas tarefas, estes aparelhos necessitam de se mover verticalmente ao longo de uma coluna de água. É neste contexto que os projetos “Deepfloat” e “Sidenav” têm como objetivo desenvolver um sistema de variação de lastro flexível para aplicações em profundidade, recorrendo ao conceito de movimento de óleo para efetuar o bombeamento da água, e, desta forma, aumentar o leque de operações a realizar em ambientes de elevadas pressões, ao mesmo tempo que se reduz a energia consumida nas mesmas e aumentando a capacidade de carga útil e de controlo em ambientes confinados. Assim, é necessário a existência de um sistema capaz de controlar a posição destes dispositivos enquanto se encontram em operações subaquáticas. Neste contexto, a presente dissertação tem como objetivo principal criar uma rotina de software que possibilite a identificação, através de técnicas de data mining, e o tracking direcional de fontes sonoras relativamente a um array de dois hidrofones, em Long Baseline (LBL), baseado na estima da Time difference of arrival (TDOA) de sinais aos mesmos. Para este efeito, são estudadas, ao longo do enquadramento teórico, matérias teóricas consideradas pertinentes para a compreensão da propagação acústica em ambiente aquático, de arquiteturas de arrays de hidrofones para o posicionamento de fontes sonoras e de identificação de sinais acústicos. Os dados utilizados para testar as rotinas em software, foram recolhidos no exercício “Robotics Exercise 2017” (REX 17), coordenado pelo Centro de Investigação Naval (CINAV). No Capítulo 8 serão discutidos os resultados obtidos, que se apresentaram com taxas de erro reduzidas. Este facto poderá dever-se à utilização de um baixo número de plataforma diferentes numa área consideravelmente reduzida.
Currently, several underwater operations performed in the open sea or rivers are carried out by dedicated systems, using normally the aid of ROV's (Remotely Operated Vehicles) and AUV's (Autonomous Underwater Vehicle). In the course of these tasks, the devices will need to move vertically along a column of water. It is in this context that the "Deepfloat" and "Sidenav" projects aim to develop a flexible ballast variation system for in-depth applications, using the concept of oil movement to pump water, increasing the range of operations to be performed in high pressure environments, while reducing the energy consumption and increasing the payload capacity and control in confined environments. Thus, it is necessary to have a system capable of controlling the position of these devices while in underwater operations. In this context, the main objective of this dissertation is to create a software routine that allows the identification, through data mining techniques, and the directional tracking of sound sources in relation to an array of two hydrophones, organized in a Long Baseline (LBL) structure, based on the estimation of Time difference of arrival (TDOA) of signals between the hydrophones. For this purpose, theoretical matters considered relevant for the understanding of acoustic propagation in the aquatic environment, of hydrophone arrays architectures used in the positioning of sound sources and the identification of acoustic signals are studied. The data used to test the software routines were collected in the "Robotics Exercise 2017" (REX 17), coordinated by Centro de Investigação Naval (CINAV). In Chapter 8 we will discuss the results obtained, which presented reduced error rates. This may be due to the use of a small number of different platforms in a considerably reduced area.
Currently, several underwater operations performed in the open sea or rivers are carried out by dedicated systems, using normally the aid of ROV's (Remotely Operated Vehicles) and AUV's (Autonomous Underwater Vehicle). In the course of these tasks, the devices will need to move vertically along a column of water. It is in this context that the "Deepfloat" and "Sidenav" projects aim to develop a flexible ballast variation system for in-depth applications, using the concept of oil movement to pump water, increasing the range of operations to be performed in high pressure environments, while reducing the energy consumption and increasing the payload capacity and control in confined environments. Thus, it is necessary to have a system capable of controlling the position of these devices while in underwater operations. In this context, the main objective of this dissertation is to create a software routine that allows the identification, through data mining techniques, and the directional tracking of sound sources in relation to an array of two hydrophones, organized in a Long Baseline (LBL) structure, based on the estimation of Time difference of arrival (TDOA) of signals between the hydrophones. For this purpose, theoretical matters considered relevant for the understanding of acoustic propagation in the aquatic environment, of hydrophone arrays architectures used in the positioning of sound sources and the identification of acoustic signals are studied. The data used to test the software routines were collected in the "Robotics Exercise 2017" (REX 17), coordinated by Centro de Investigação Naval (CINAV). In Chapter 8 we will discuss the results obtained, which presented reduced error rates. This may be due to the use of a small number of different platforms in a considerably reduced area.
Description
Keywords
Time difference of arrival, Data mining, Tracking direcional, Processamento de sinais, Array de hidrofones Time difference of arrival, Data mining, Directional Tracking, Signal Processing, Hydrophone array