基於拖曳式水下聽音陣列之水下定位研究

Abstract

拖曳式水下聽音陣列（Towed Hydrophone Array）為一種具高機動性與廣域覆蓋能力的被動聲學監測平台。透過船舶拖曳，即時接收水下聲源訊號並進行方向與位置估算。本研究聚焦於拖曳式水下聽音陣列於水下目標定位之實際應用，涵蓋海難搜救與海洋哺乳類生態監測兩大情境，驗證其定位效能。 首先，於海空難搜救應用中，本研究針對8.8 kHz DK180水下發報器訊號，採用到達時間差法（Time Difference of Arrival, TDOA）與網格疊加定位技術實現近即時的聲源定位，並據此調整偵蒐船航向，模擬實際搜索場景中的動態作業流程。透過在小琉球外海進行之實海域實驗，進行多種海流與船速條件下之測試，並納入音傳損耗與拖曳式水下聽音陣列姿態修正之後處理，有效提升定位準確性與搜尋效率，為未來飛航紀錄器回收等搜救任務提供實用的技術支持與作業標準。此項成果顯示在水深480公尺以內可達到高精度的水下定位，未來可在深海域進行測試，俾能在航空事故救援中廣泛應用，進一步提高事故調查的效率和準確性。 生態監測應用則利用DCLDE（Detection, Classification, Localization, and Density Estimation, DCLDE）資料集進行抹香鯨(Physeter macrocephalus)之搭聲（Click）偵測與三維定位。透過建構三維網格，結合TDOA與斜距延遲分析技術，實現深潛物種之深度與行進方向推測。此技術除有望於未來結合即時演算法與聲紋辨識技術，建立更完整的海洋哺乳動物行為資料庫。 本研究建立了一套適用於不同水下目標定位需求的拖曳式聲學監測系統，不僅在實務測試中展現即時性與高機動優勢，亦具備拓展至長期生態監測與跨物種辨識的潛力，為海洋科學研究與水下搜救行動提供一套兼具效率與準確性的技術平台。

The towed hydrophone array is a passive acoustic monitoring platform characterized by high mobility and broad coverage capabilities. By towing an array of hydrophones behind a vessel, it enables real-time reception of underwater acoustic signals and estimation of the direction and position of sound sources. This study focuses on the practical application of towed arrays in underwater target localization, specifically in two key scenarios: maritime search and rescue, and marine mammal ecological monitoring, in order to validate the system’s localization performance. In the context of search and rescue operations, this study targets the 8.8 kHz signal from DK180 underwater beacons, applying the Time Difference of Arrival (TDOA) method and grid-based localization techniques to achieve near real-time sound source positioning. The estimated direction is then used to dynamically adjust the search vessel’s heading, simulating real-world search operations. Field experiments conducted off the coast of Liuqiu Island tested the system under various ocean current and vessel speed conditions. Post-processing procedures including acoustic propagation loss compensation and array attitude correction were incorporated to enhance localization accuracy and search efficiency. The results demonstrated that high-precision underwater localization is achievable at depths up to 480 meters. Future deep-sea trials are planned to expand the system’s application in aviation accident recovery, potentially improving the effectiveness and accuracy of accident investigations. In the ecological monitoring application, the study utilizes the Detection, Classification, Localization, and Density Estimation (DCLDE) dataset to detect and localize the clicks of sperm whales (Physeter macrocephalus) in three dimensions. By constructing a 3D localization grid and combining TDOA with slant range delay analysis, the system effectively estimates the depth and movement direction of deep-diving marine mammals. This technique shows promise for integration with real-time algorithms and acoustic fingerprinting technologies, supporting the development of a comprehensive behavioral database for marine mammals. Overall, this research presents a versatile towed acoustic monitoring system tailored to various underwater localization needs. Demonstrating both real-time capability and operational flexibility in field trials, the system also offers strong potential for expansion into long-term ecological monitoring and multi-species identification. It provides a robust and accurate technical platform for marine scientific research and underwater search and rescue missions.