AN IA-ORIENTED POULTRY FARM MONITORING SYSTEM FOR REAL-TIME ENVIRONMENTAL RISK ASSESSMENT AND ALERT MANAGEMENT

Abstract

Poultry farming is an important component of agricultural production and food security; however, effective management of poultry houses requires continuous monitoring of environmental and production conditions. Conventional monitoring practices often depend on manual observation and record keeping, which can result in delayed identification of abnormal conditions, inconsistent records, and slow responses to potentially harmful environmental changes. This study presents the design and development of an IoT-oriented framework for environmental monitoring and risk assessment in poultry farm management. The study adopted a hybrid Object-Oriented Analysis and Design (OOAD) and iterative prototyping methodology to develop a functional web-based poultry farm monitoring system. The proposed system was structured into four layers: data acquisition, communication and data storage, processing, and application/presentation. The system provides facilities for poultry-house registration, batch management, sensor-reading entry, configurable threshold management, environmental risk assessment, dashboard visualization, historical record management, role-based access control, and automated alert generation. Eight poultry-related indicators are considered, including temperature, humidity, ammonia level, light intensity, feed level, water level, mortality count, and bird-weight sample. A centralized MySQL relational database was implemented to store monitoring records, while a rule-based risk assessment engine evaluates environmental readings against administrator-configured threshold values and classifies conditions as Normal, Warning, or Critical. Where abnormal conditions are detected, corresponding alert records are generated and email notification is attempted. System testing covered critical environmental readings, invalid batch selection, dynamic batch retrieval, threshold reconfiguration, duplicate user registration, and role-restricted access. The test results showed that the major functional components operated as expected after identified implementation defects were corrected. In particular, a test temperature reading of 40°C and humidity reading of 75% successfully triggered Critical and Warning classifications respectively, demonstrating the effectiveness of the implemented threshold-based assessment mechanism. The study demonstrates that the proposed framework can provide a structured digital approach to poultry farm monitoring, improve data organization, support consistent environmental risk classification, and facilitate timely notification of abnormal conditions. However, the current implementation uses simulated sensor inputs and a rule-based assessment mechanism rather than physically deployed IoT sensors and a trained machine-learning model. The developed framework therefore provides a practical foundation for future integration of physical sensing devices, machine-learning algorithms, predictive analytics, automated actuators, and mobile-based remote monitoring for intelligent poultry farm management.