Pengembangan Sistem Otomatisasi Nutrisi Hidroponik Berbasis Internet of Things dengan Logika Fuzzy

Abstract

Penelitian ini mengembangkan sistem otomatisasi pemberian nutrisi pada budidaya selada keriting hidroponik berbasis Internet of Things (IoT) dengan logika fuzzy Mamdani untuk mengatasi keterbatasan pengecekan manual dan mendukung pemantauan jarak jauh. Perangkat dirancang menggunakan mikrokontroler ESP32, sensor Total Dissolved Solids (TDS) untuk mengukur konsentrasi nutrisi, dan sensor DS18B20 untuk suhu larutan. Data dikirim ke aplikasi Blynk dan basis data MySQL setiap 10 detik. Fungsi keanggotaan dan aturan fuzzy diuji serta divalidasi melalui perbandingan perhitungan di MATLAB. Uji selama tiga hari menunjukkan akurasi rata rata sensor TDS 96,84% (MAE 16?ppm; RMSE 17,73?ppm) dan DS18B20 97,26% (MAE 0,46?°C; RMSE 0,54?°C). Durasi kerja pompa hasil inferensi fuzzy hanya berbeda rata rata 0,024?detik (2,4%) dibanding MATLAB (MAE 0,02?detik; RMSE 0,0447?detik). Sistem mampu mempertahankan nutrisi dalam rentang optimal dan menyediakan pemantauan secara jarak jauh.

This study presents the development of an Internet of Things (IoT)–based automatic nutrient dosing system for hydroponic cultivation of curly lettuce, employing Mamdani fuzzy logic to overcome the limitations of manual monitoring and to enable remote supervision. The device is built around an ESP32 microcontroller and incorporates a Total Dissolved Solids (TDS) sensor to measure nutrient concentration and a DS18B20 sensor to monitor solution temperature. Data are transmitted to a Blynk application and stored in a MySQL database every 10 seconds. The fuzzy membership functions and rule base were tested and validated by comparing calculations performed in MATLAB. A three-day trial demonstrated an average TDS sensor accuracy of 96.84?% (MAE 16?ppm; RMSE 17.73?ppm) and a DS18B20 accuracy of 97.26?% (MAE 0.46?°C; RMSE 0.54?°C). The pump activation durations determined by the fuzzy inference system differed by an average of only 0.024 seconds (2.4?%) from MATLAB results (MAE 0.02?s; RMSE 0.0447?s). The system successfully maintained nutrient levels within the optimal range and provided reliable remote monitoring.