Genetic algorithm-based chicken manure weight prediction system development

Abstrak

This research presents design and implementation of internet of things (IoT) based monitoring and predictive system for evaluating chicken manure weight and environmental conditions in poultry housing. The proposed system integrates MQ-137 sensor for ammonia detection, DHT22 sensor for temperature and humidity measurement, and load cell modules for manure weight monitoring. All sensor data are transmitted in real time to cloud platform, enabling continuous environmental assessment. A 30-day experimental study was conducted using two controlled chicken drum models, each containing 15 broiler chickens and provided with different feed types to observe variations in manure production and air quality. Sensor calibration results indicate high accuracy, with average error of 0.31% for ammonia readings and 0.10% for manure weight measurement. Experimental findings show that feed type A generates lower manure weight, reduced ammonia concentration, and more stable temperature conditions compared to feed type B, suggesting improved feed efficiency and better overall chicken health. A genetic algorithm (GA) was employed to optimize regression model predicting manure weight using ammonia concentration and temperature as input features. The GA-optimized model achieved strong predictive performance, with root mean square error (RMSE) of 0.358 g and coefficient of determination (R2) value of 0.992. The results demonstrate that proposed system provides reliable, scalable, and data-driven solution for smart poultry monitoring and early health detection.