Smart Farm Automation for Optimized Poultry Housing Climate Control

Thesis Abstract

The rapidly evolving convergence of Internet of Things (IoT), sensor networks, and data analytics offers transformative potential for improving welfare, productivity, and sustainability in poultry production, yet conventional housing systems remain prone to suboptimal microclimate management, disease risk, and resource waste due to delayed or non-digital decision processes. This study addresses the gap by developing and evaluating an integrated smart farm automation system that dynamically optimizes poultry housing climate control to enhance performance, welfare, and energy efficiency. The aim is to design, deploy, and validate an ICT-driven climate management framework that integrates environmental sensing, predictive analytics, and automated actuators within commercial broiler houses. Specific objectives include (i) characterizing ambient microclimate dynamics and their relationships with growth and welfare indicators; (ii) developing a sensor-driven control architecture incorporating temperature, humidity, CO2, ammonia, and airflow data streams; (iii) applying predictive models to forecast thermal stress and odor levels and to generate optimal setpoints; (iv) implementing automated ventilation, heating, humidification, and curtain actuators with fail-safe and energy-saving strategies; and (v) evaluating system performance in terms of animal growth, mortality, feed conversion ratio, welfare indicators, and energy consumption. A mixed-methods research design will be employed in two phases. Phase I involves a quantitative field study in two commercial broiler houses (n = 2) over 12 production cycles, with a total sample of 12,000 birds per house. Data collection will utilize calibrated environmental sensors (temperature, relative humidity, CO2, NH3, PM2.5), automated climate control logs, and growth metrics (weight gain, feed intake, mortality). Phase II comprises a quasi-experimental intervention in a matched pair of houses (n = 2) over six weeks, comparing the smart control treatment against conventional practice. Instruments include validated welfare assessment checklists, a farmer-completed usability survey, and system reliability logs. Validity and reliability will be ensured through sensor cross-validation, pilot testing (n = 200 birds per house), and triangulation of management records with automated data streams. Analytical methods will encompass time-series analysis and regression to quantify relationships between microclimate variables and performance outcomes, ANOVA to compare treatment effects on growth and feed efficiency, and multivariate modeling (partial least squares structural equation modeling) to capture causal pathways among climate, welfare indicators, and production metrics. Predictive modeling will deploy recurrent neural networks (LSTM) to forecast thermal stress and odor events, with model evaluation via RMSE and R-squared metrics. The conceptual framework will be anchored in the biosecurity and welfare theories of environmental fit and the Technology Acceptance Model (TAM) to investigate adoption determinants among farm staff. The study will also conduct a cost-benefit analysis (CBA) to quantify economic implications, including energy savings and reduced mortality. Expected findings include (i) improved microclimate stability (reduced diurnal temperature variation and more consistent CO2 and NH3 levels), (ii) statistically significant improvements in average daily gain, feed conversion ratio, and mortality reductions relative to baseline, (iii) enhanced welfare scores through optimized stocking density and ventilation patterns, and (iv) measurable reductions in energy use and greenhouse gas emissions due to adaptive control strategies. The study contributes to knowledge by presenting an integrative ICT-driven framework for poultry housing climate management, demonstrating how real-time sensing, predictive analytics, and automated actuation co-create productive and welfare-enhancing environments. It also advances methodical understanding of how digital technologies can be harmonized with farm operations to achieve sustainability goals, including a demonstration of LSTM-based forecasting in an agricultural automation context. The main conclusion is that tightly integrated smart climate control can outperform traditional methods in balancing productive performance, animal welfare, and energy efficiency in commercial poultry housing. Recommendations include scaling the system to multi-house farms with modular architectures, enhancing data governance and cybersecurity, and tailoring user interfaces to farmer workflows to promote sustained adoption. Further research is suggested to explore Cloud-edge hybrid architectures for latency-sensitive control, integration with disease monitoring for proactive biosecurity, and longitudinal assessments across different poultry species and climatic zones.