Predictive Modelling of Egg Production Yields on Farms based on Environmental Factors

Abstract

This research presents an integrated smart farming system aimed at optimizing egg yield on poultry farms by leveraging artificial intelligence (AI), Internet of Things (IoT), and environmental sensing technologies. The system is structured around four core components - Animal Stress Monitoring, Temperature Control and Predator Detection, Humidity and Ventilation Management, and AI-Driven Smart Lighting Optimization each contributing to real-time environmental adaptation and accurate egg production prediction. Animal stress is assessed using physiological and environmental metrics (e.g., heart rate, body temperature, feed/water intake), with predictions generated via an XGBoost model trained on 3000+ real farm entries. Temperature and security are managed through a hybrid system combining DHT11/DHT22-based climate control with YOLO-based computer vision for predator detection. The humidity and ventilation module incorporates Bi-LSTM and XGBoost models to predict and regulate airflow and moisture levels based on real-time sensor inputs. The lighting optimization component dynamically adjusts LED spectrum and intensity using LSTM-based forecasting models, operating via ESP32 and MQTT-enabled architecture to simulate ideal lighting conditions. These components are unified through a.NET-based backend and a mobile-friendly dashboard, enabling low-latency decision support and seamless farm management. The system's modularity, edge deployment capabilities, and adaptability to local conditions make it an innovative and scalable approach for enhancing egg yield, poultry welfare, and farm automation.