Sistem Pemeliharaan dan Penyiraman Tanaman Pot Berbasis IoT dan Machine Learning dengan Rekomendasi Perawatan Adaptif

Muhammad Hidayatul Arifin (State University of Medan)
Dedy Kiswanto (State University of Medan)
Muhammad Raffi akbar Tanjung (State University of Medan)
Ririn Amelia Br Siregar (State University of Medan)

Jurnal Nasional Komputasi dan Teknologi Informasi
Vol. 9 No. 1 (2026): Februari, 2026

Abstrak - Metode irigasi konvensional sering kali kurang responsif terhadap dinamika cuaca mikro, yang berisiko mengganggu pertumbuhan tanaman. Penelitian ini merancang sistem irigasi otonom berbasis server (VPS) yang memadukan Internet of Things (IoT) dengan Hybrid Artificial Intelligence. Menggantikan mekanisme timer statis, sistem ini memanfaatkan algoritma Decision Tree untuk menghasilkan keputusan adaptif, yang dilatih menggunakan 3.000 data sintetis berbasis aturan pakar. Evaluasi ketat melalui 5-Fold Cross-Validation membuktikan stabilitas sistem yang tinggi, dengan rata-rata akurasi diagnosis mencapai 99,97% dan kesalahan prediksi volume air (Mean Absolute Error) hanya 1,02 ml. Validasi eksternal pada perangkat keras mengonfirmasi bahwa fitur cerdas, seperti penundaan penyiraman saat suhu ekstrem (Smart Delay) dan mode malam (Night Mode), berfungsi secara presisi di lingkungan nyata. Studi ini menyimpulkan bahwa integrasi AI terpusat efektif mewujudkan manajemen pertanian presisi yang handal dan adaptif. Kata kunci : Internet of Things, Machine Learning, Perawatan Tanaman, Smart Farming, Irigrasi Tanaman. Abstract - Conventional irrigation often fails to respond effectively to microclimatic changes, posing risks to plant health. This study designs an autonomous irrigation system hosted on a VPS, integrating IoT with Hybrid Artificial Intelligence. Replacing static timers, the system employs Decision Tree algorithms for adaptive decision-making, trained on 3,000 expert-rule-based synthetic data points. Rigorous evaluation using 5-Fold Cross-Validation demonstrated high stability, achieving an average diagnostic accuracy of 99.97% and a water volume prediction error (Mean Absolute Error) of just 1.02 ml. External validation on hardware confirmed that intelligent features, such as Smart Delay during extreme heat and Night Mode, function precisely in real-world environments. This study concludes that centralized AI integration effectively realizes reliable and adaptive precision agriculture management. Keywords: Internet of Things, Machine Learning, Plant Care, Smart Farming, Plant Irrigation.