Implementation of Bidirectional Long Short-Term Memory (BiLSTM) to Predict the Speed and Direction of Sea Level Flows in the Bali Strait Shipping Lane

Sumanto, Alvico Faudiansyah (2026) Implementation of Bidirectional Long Short-Term Memory (BiLSTM) to Predict the Speed and Direction of Sea Level Flows in the Bali Strait Shipping Lane. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (COVER) Cover Skripsi.pdf Download (1MB)
	Text (BAB 1) CHAPTER I INTRODUCTION .pdf Download (171kB)
	Text (BAB 2) CHAPTER II LITERATURE REVIEW .pdf Restricted to Repository staff only until 18 July 2029. Download (507kB)
	Text (BAB 3) CHAPTER III RESEARCH METHODOLOGY .pdf Restricted to Repository staff only until 18 July 2029. Download (363kB)
	Text (BAB 4) CHAPTER IV RESULTS AND DISCUSSION .pdf Restricted to Repository staff only until 18 July 2029. Download (820kB)
	Text (BAB 5) CHAPTER V CONCLUSIONS AND SUGGESTIONS.pdf Download (153kB)
	Text (DAFTAR PUSTAKA) BIBLIOGRAPHY.pdf Download (140kB)

Abstract

The Bali Strait is a strategic shipping lane with very high traffic intensity, reaching more than 85 thousand ship trips and serving 9.2 million passengers per year. The unique oceanographic conditions of the Bali Strait, namely the topography resembling a funnel, the influence of cross-Indonesian currents (Arlindo), tidal cycles, and monsoon wind variations, produce fluctuating, strong, and non-linear movements of sea surface currents that are very difficult to predict using conventional hydrodynamics models. This study implements the Bidirectional Long Short-Term Memory (BiLSTM) method as a data-driven approach to predict the speed and direction of sea surface currents in the Bali Strait. The data used is in the form of daily time series data of the east-west (U) and north-south (V) flow speed components obtained from the Copernicus Marine Service (CMEMS) with a time range of November 22, 2022 to January 22, 2026, totaling 16,213 data. Hyperparameter tests were carried out on hidden layer (50, 100, 150) and batch size (32, 64, 128, 256) variations with a fixed number of epochs of 50. The results showed that the best configuration was achieved at hidden layer 50 and batch size 32, with the MAPE value of component V being 12.03% (accurate category). In the U component, the MAPE value is not defined (inf) due to the presence of three zero-value data points that are divisors in the MAPE formula, even though visually the prediction is able to follow the actual trend. Validation using the last 9 days of the dataset (23–31 January 2026) shows that the MAPE speed in the first 5 days is below 2.5% with a best value of 1.91%, as well as a very low directional MAPE ranging from 0.34% to 4.00%. The cardinal cardinal conversion shows a full match between the prediction and the actual, i.e. the direction of the East (T) for January 23–28, 2026 and the Northeast (TL) for January 29–31, 2026. These results prove that the BiLSTM model has good generalization capabilities and has the potential to be applied as a navigation decision support system in the waters of the Bali Strait.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Haromainy, Muhammad Muharrom Al	NIDN0701069503	muhammad.muharrom.if@upnjatim.ac.id
Thesis advisor	Nurlaili, Afina Lina	NIDN0013129303	afina.lina.if@upnjatim.ac.id

Subjects:

Q Science > QA Mathematics > QA76.87 Neural computers

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Mr Alvico Faudiansyah Sumanto

Date Deposited:

20 Jul 2026 03:28

Last Modified:

20 Jul 2026 06:02

URI:

https://repository.upnjatim.ac.id/id/eprint/56232