EVALUASI KINERJA ALGORITMA KONTROL PADA SISTEM IRIGASI CERDAS

Arya Muhammad Dendyana, - and Ade Gafar Abdullah, - (2025) EVALUASI KINERJA ALGORITMA KONTROL PADA SISTEM IRIGASI CERDAS. S1 thesis, Universitas Pendidikan Indonesia.

Abstract

Penelitian ini mengevaluasi kinerja berbagai algoritma kontrol berbasis machine learning untuk prediksi evapotranspirasi referensi (ETo) sebagai dasar pengendalian sistem irigasi cerdas. Data meteorologi diperoleh dari Google Earth Engine (ERA5-Land dan FAO WaPOR) kemudian diproses dan digunakan untuk melatih delapan model regresi (LR, SVR, KNR, RFR, BR, ABR, GBR, XGBR) serta Adaptive Neuro-Fuzzy Inference System (ANFIS). Hasil evaluasi menunjukkan bahwa Support Vector Regression (SVR) adalah model terbaik pada skenario lengkap dengan nilai R² = 0.8088 dan RMSE = 2.36 mm/hari, sedangkan Linear Regression (LR) paling sesuai untuk implementasi nyata dengan sensor terbatas (R²=0.5949). Model LR diimplementasikan pada mikrokontroler ESP32 dan diuji di lapangan menggunakan sensor kelembapan tanah dan sensor DHT11. Pengujian nyata menunjukkan bahwa metode LR mampu memberikan volume air lebih sesuai dengan kebutuhan evapotranspirasi dibandingkan metode threshold konvensional, sehingga lebih efisien dan mendukung pertumbuhan tanaman secara optimal. Hasil ini membuktikan bahwa integrasi machine learning dengan perangkat IoT berbiaya rendah dapat menjadi solusi praktis untuk sistem irigasi presisi. Kata Kunci: Smart Irrigation, ESP32, MQTT, Fuzzy Logic, Machine Learning, IoT The increasing demand for food due to population growth and climate change demands innovation for more efficient irrigation systems. This research evaluates the performance of various machine learning-based control algorithms for predicting reference evapotranspiration (ETo) as a basis for controlling a smart irrigation system. Meteorological data was obtained from Google Earth Engine (ERA5-Land and FAO WaPOR), then processed and used to train eight regression models (LR, SVR, KNR, RFR, BR, ABR, GBR, XGBR) and an Adaptive Neuro-Fuzzy Inference System (ANFIS). The evaluation results show that Support Vector Regression (SVR) was the best model in the full-feature scenario, with an R² of 0.8088 and an RMSE of 2.36 mm/day, while Linear Regression (LR) was most suitable for real-world implementation with limited sensors (R² = 0.5949). The LR model was implemented on an ESP32 microcontroller and field-tested using a soil moisture sensor and a DHT11 sensor. Field testing demonstrated that the LR method could provide a water volume more aligned with evapotranspiration needs compared to the conventional threshold method, thus being more efficient and supporting optimal plant growth. These results prove that integrating machine learning with low-cost IoT devices can be a practical solution for precision irrigation systems. Keywords: Smart Irrigation, Machine Learning, Evapotranspirasi, ESP32, Linear Regression

	Text S_TE_2104306_Chapter 1.pdf Download (9MB)
	Text S_TE_2104306_Chapter 2.pdf Restricted to Staf Perpustakaan Download (11MB)
	Text S_TE_2104306_Chapter 3.pdf Download (33MB)
	Text S_TE_2104306_Chapter 4.pdf Restricted to Staf Perpustakaan Download (19MB)
	Text S_TE_2104306_Chapter 5.pdf Download (3MB)
	Text S_TE_2104306_Appendix.pdf Restricted to Staf Perpustakaan Download (2MB)
	Text S_TE_2104306_itle.pdf Download (1MB)

Item Type:	Thesis (S1)
Additional Information:	SINTA ID DOSEN PEMBIMBING Ade Gafar Abdullah: 257412
Uncontrolled Keywords:	Smart Irrigation, ESP32, MQTT, Fuzzy Logic, Machine Learning, IoT
Subjects:	L Education > L Education (General) Q Science > QA Mathematics > QA76 Computer software T Technology > T Technology (General)
Divisions:	Fakultas Pendidikan Teknik dan Industri > Jurusan Pendidikan Teknik Elektro
Depositing User:	Arya Muhammad Dendyana
Date Deposited:	31 Oct 2025 15:00
Last Modified:	31 Oct 2025 15:00
URI:	http://repository.upi.edu/id/eprint/144122

Actions (login required)

View Item