Pemodelan Deteksi Kehilangan Air Pada Jaringan Distribusi Air Berbasis Jaringan Syaraf Tiruan di Perumda Air Minum Tugu Tirta Malang

Nigel, Matheo (2023) Pemodelan Deteksi Kehilangan Air Pada Jaringan Distribusi Air Berbasis Jaringan Syaraf Tiruan di Perumda Air Minum Tugu Tirta Malang. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 02311940000073_Undergraduate-Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2025. Download (1MB) | Request a copy

Abstract

Kehilangan air masih sering terjadi pada jaringan distribusi air, terutama pada Perumda yang menaungi banyak masyarakat. Oleh karena itu, penelitian ini menggunakan Jaringan Syaraf Tiruan sebagai opsi metode deteksi kehilangan air dengan memprediksi aliran debit yang tercatat pada datalogger DMA (District Meter Area) tertentu. Algoritma JST yang digunakan sendiri menggunakan Long Short-Term Memory atau LSTM untuk memprediksi hasil aliran debit pada DMA yang dianalisis. Dalam penelitian tugas akhir ini, setelah melalui studi literatur dan lapangan, dilakukan validasi model menggunakan beberapa iterasi sampai algoritma mencapai nilai mendekati optimal. Setelah itu, algoritma LSTM digunakan untuk memprediksi aliran debit selama beberapa hari. Hasil menunjukkan untuk performansi algoritma pada DMA MOJO 3B2 memiliki nilai Mean Absolute Percentage Error sebesar 16,4% dengan Root Mean-Squared Error sebesar 2,92 dan untuk performansi algoritma pada DMA WENDIT 1N memiliki nilai Mean Absolute Percentage Error sebesar 12,6% dengan Root Mean-Squared Error sebesar 1,75. Kemudian untuk perbedaan kehilangan air data simulasi dengan aktual sendiri memiliki rerata perbedaan sebesar 0,57% untuk DMA MOJO 3B2 dan 0,93% untuk DMA WENDIT 1N. Dengan nilai kehilangan air dengan perbedaan kurang dari 1%, maka dapat disimpulkan bahwa algoritma dari JST dapat digunakan untuk mendeteksi adanya kehilangan air pada jaringan distribusi air.
===============================================================================================================================
Water loss is still common in water distribution networks, especially in Perumda which houses many communities. Therefore, this study uses neural network as an option method of detecting water loss by predicting the flow of discharge recorded on a particular DMA datalogger. The neural network algorithm used itself uses LSTM to predict the discharge flow results on the analyzed DMA. In this final project research, after going through literature and field studies, model validation was carried out using several iterations until the algorithm reached a value close to optimal. After that, the LSTM algorithm is used to predict the flow of discharge over several days. The results show that for algorithm performance in DMA MOJO 3B2 has a Mean Absolute Percentage Error value of 16.4% with a Root Mean-Squared Error of 2.92 and for algorithm performance in DMA WENDIT 1N has a Mean Absolute Percentage Error value of 12.6% with a Root Mean-Squared Error amounted to 1.75. Then for the difference in water loss, the simulation data with the actual itself has an average difference of 0.57% for DMA MOJO 3B2 and 0.93% for DMA WENDIT 1N. With a water loss value with a difference of less than 1%, it can be concluded that the algorithm of neural network can be used to detect water loss in water distribution networks.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Aliran Debit, District Meter Area, Jaringan Syaraf Tiruan, Kehilangan Air, Artificial neural network, Discharge flow, District Meter Area, Water Loss
Subjects:	Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) T Technology > TD Environmental technology. Sanitary engineering > TD481 Water distribution systems
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Matheo Nigel
Date Deposited:	29 Aug 2023 02:13
Last Modified:	29 Aug 2023 02:13
URI:	http://repository.its.ac.id/id/eprint/100664

Actions (login required)

View Item