SISTEM MONITORING KUALITAS AIR DAN PREDIKSI SKOR STRES IKAN LELE (Clarias sp.) 24 JAM KE DEPAN BERBASIS INTERNET OF THINGS MENGGUNAKAN JARINGAN SARAF TIRUAN DENGAN PARAMETER AMONIA DAN SUHU

Abstract

Ammonia concentration and water temperature are two critical water-quality parameters in intensive catfish (Clarias sp.) farming, since both can trigger physiological stress and reduce productivity when monitoring is infrequent. Conventional monitoring is generally still performed manually with limited measurement frequency, making it prone to delayed detection of critical conditions. This study designed and built a web-based Internet of Things (IoT) monitoring system that integrates a DFRobot SEN0469 ammonia sensor and a DFRobot DS18B20 temperature sensor with an Arduino Portenta H7 microcontroller, transmitting data through the ThingSpeak IoT platform to a real-time web dashboard. Field testing was conducted over 8 days (20-28 May 2026) on a small-scale circular tarpaulin pond, collecting 20,269 raw data points that were resampled into 194 hourly points. Sensor accuracy verification against calibrated reference instruments (API ammonia test kit and a calibrated thermometer) on five comparison points showed a Mean Absolute Error (MAE) of 0.024 ppm for the ammonia sensor and 0.062°C for the temperature sensor, with the temperature sensor further verified for precision, linearity (R² = 0.999), sensitivity, and resolution. During the observation period, ammonia concentration generally remained within the safe category (median 0.016 ppm), while temperature showed larger fluctuations relative to the optimal range, and correlation analysis indicated that temperature contributed more dominantly to fluctuations in the fish's water-quality-based condition than ammonia. The results demonstrate that the developed sensor-microcontroller-IoT dashboard pipeline reliably acquires, transmits, and displays real-time water-quality data, providing a practical monitoring and early-warning tool for small-scale catfish farmers.