<![CDATA[Abstrak - Metode irigasi konvensional sering kali memiliki keterbatasan dalam merespons dinamika mikroklimat, yang berpotensi mengganggu pertumbuhan tanaman. Penelitian ini bertujuan merancang sistem pemeliharaan tanaman otonom yang memadukan teknologi Internet of Things (IoT) dengan Hybrid Artificial Intelligence dalam kerangka arsitektur server terpusat (VPS). Berbeda dengan sistem berbasis penjadwalan waktu (timer), solusi ini menerapkan algoritma Decision Tree untuk menghasilkan keputusan adaptif yang didasarkan pada pembacaan sensor secara real-time. Pendekatan metodologi mencakup pembangkitan 3.000 data latih sintetis berdasarkan logika pakar guna melatih model dalam menangani berbagai skenario kompleks, termasuk kondisi kekeringan ekstrem dan anomali pencahayaan. Hasil evaluasi komputasional menunjukkan kinerja sistem yang presisi, dengan akurasi diagnosis mencapai 100% dan rata-rata kesalahan prediksi volume air (Mean Absolute Error) hanya sebesar 0,66 ml. Lebih lanjut, pengujian fungsional pada perangkat keras memvalidasi stabilitas dan responsivitas sistem dalam menjalankan logika kendali cerdas, seperti penundaan penyiraman saat suhu tinggi (Smart Delay) dan penguncian pompa otomatis pada malam hari (Night Mode). Studi ini menyimpulkan bahwa integrasi AI dalam arsitektur IoT terpusat efektif untuk mewujudkan sistem manajemen tanaman yang presisi, adaptif terhadap lingkungan, dan handal untuk implementasi praktis.Kata kunci : Internet of Things; Machine Learning; Perawatan Tanaman; Smart Farming; Irigrasi Tanaman; Abstract - Conventional irrigation methods often have limitations in responding to microclimate dynamics, potentially disrupting plant growth. This study aims to design an autonomous plant maintenance system that combines Internet of Things (IoT) technology with Hybrid Artificial Intelligence within a centralized server (VPS) architecture. Unlike timer-based systems, this solution applies a Decision Tree algorithm to generate adaptive decisions based on real-time sensor readings. The methodological approach includes generating 3,000 synthetic training data sets based on expert logic to train the model to handle various complex scenarios, including extreme drought conditions and lighting anomalies. Computational evaluation results demonstrate precise system performance, with a diagnostic accuracy reaching 100% and an average water volume prediction error (Mean Absolute Error) of only 0.66 ml. Furthermore, functional testing on hardware validates the system's stability and responsiveness in executing intelligent control logic, such as delaying watering during high temperatures (Smart Delay) and automatic pump locking at night (Night Mode). This study concludes that the integration of AI within a centralized IoT architecture is effective in realizing a precise, environmentally adaptive, and reliable plant management system for practical implementation.Keywords: Internet of Things; Machine Learning; Plant Care; Smart Farming; Plant Irrigation;]]>
Copyrights © 2026
