Deteksi Objek Berdasarkan Sinyal Akustik Bawah Air Dari Hydrophone Array Dengan Menggunakan Metode Machine Learning

Pramono, Muhammad Firdaus (2026) Deteksi Objek Berdasarkan Sinyal Akustik Bawah Air Dari Hydrophone Array Dengan Menggunakan Metode Machine Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5048221024-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Deteksi objek dibawah air merupakan salah satu aspek penting dalam pengembangan sistem pemantauan dan pengawasan lingkungan laut. Di lingkungan bawah air, gelombang elektromagnetik tidak dapat merambat dengan baik sehingga pendeteksian objek lebih efektif dilakukan dengan menggunakan sinyal akustik. Penelitian ini bertujuan untuk mengembangkan sistem deteksi dan klasifikasi objek berdasarkan sinyal akustik bawah air yang direkam dengan menggunakan hydrophone array, dengan menerapkan metode machine learning sebagai pendekatan utama. Data akustik diperoleh dari pengujian di kolam uji dengan tiga kelas data, yaitu penyelam (Diver), model kapal (Ship Model), dan tanpa objek (No Object). Sinyal yang didapatkan diproses melalui tahap filtering, segmentasi sinyal, serta ekstraksi fitur menggunakan metode Power Spectral Density (PSD) untuk mendapatkan nilai rata-rata PSD yang merepresentasikan karakteristik masing-masing kelas. Kemudian, Fitur PSD digunakan sebagai masukan pada metode Support Vector Machine (SVM) untuk melakukan klasifikasi objek. Hasil Pengujian menunjukkan bahwa metode PSD dan SVM mampu membedakan karakteristik akustik antara kelas Diver, Ship Model, dan No Object dengan baik. Pada Klasifikasi Diver dan No Object mendapatkan akurasi pengujian sebesar 92,41%, sedangkan Pada Klasifikasi Ship Model dan No Object mendapatkan akurasi pengujian sebesar 100%. Sehingga pengujian ini dapat digunakan sebagai pendekatan dalam sistem deteksi objek bawah air berbasis sinyal akustik bawah air.
========================================================================================================================================
Underwater object detection is an important aspect in the development of marine environmental monitoring and surveillance systems. In underwater environments, electromagnetic waves cannot propagate well, so object detection is more effective using acoustic signals. This study aims to develop an object detection and classification system based on underwater acoustic signals recorded using a hydrophone array, by applying machine learning methods as the main approach. Acoustic data was obtained from testing in a test pool with three data classes, namely divers (Diver), ship models (Ship Model), and no objects (No Object). The obtained signals were processed through filtering, signal segmentation, and feature extraction stages using the Power Spectral Density (PSD) method to obtain the average PSD value that represents the characteristics of each class. Then, the PSD features were used as input to the Support Vector Machine (SVM) method to classify objects. Test results show that the PSD and SVM methods are able to distinguish the acoustic characteristics between the Diver, Ship Model, and No Object classes well. The Diver and No Object classifications achieved a test accuracy of 92.41%, while the Ship Model and No Object classifications achieved a test accuracy of 100%. Therefore, this test can be used as an approach in underwater object detection systems based on underwater acoustic signals

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Akustik bawah air, Deteksi Objek, Hydrophone, Power Spectral Density (PSD), Support Vector Machine (SVM), Object Detection, Underwater Acoustics
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5101 Telecommunication
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Telecommunication Engineering > 20202-(S1) Undergraduate Thesis
Depositing User:	Muhammad Firdaus Pramono
Date Deposited:	24 Jul 2026 01:00
Last Modified:	24 Jul 2026 01:00
URI:	http://repository.its.ac.id/id/eprint/136840

Actions (login required)

View Item