This master’s thesis, authored by Zhou Zhou at Cyprus University of Technology in June 2025, explores the prediction of indoor air quality (IAQ) and the development of improvement strategies using the ETSformer neural network. The work is situated within the Department of Electrical Engineering, Computer Engineering, and Informatics, reflecting a multidisciplinary approach that combines environmental science, machine learning, and engineering. The thesis addresses the growing need for accurate IAQ forecasting and actionable strategies to mitigate air pollution in indoor environments, which is critical for public health and environmental management.
Application of ETSformer Neural Network: The thesis leverages the ETSformer neural network architecture, a state-of-the-art model for time series prediction, to forecast IAQ levels. This approach builds on recent advances in transformer-based models, which have demonstrated superior performance in capturing complex spatiotemporal dependencies in environmental data.
Development of IAQ Improvement Strategies: Beyond prediction, the thesis proposes practical strategies for improving IAQ based on model outputs. These strategies are tailored to real-world scenarios, considering both technical feasibility and potential impact on occupant health and comfort.
Integration of Data Sources: The research integrates multiple data streams, including sensor measurements and contextual information, to enhance the robustness and accuracy of IAQ predictions. This holistic data-driven methodology enables more reliable decision-making for building management and policy interventions.
Evaluation and Validation: The thesis includes a comprehensive evaluation of the ETSformer model’s predictive performance, benchmarking it against existing methods. The results demonstrate notable improvements in forecasting accuracy, underscoring the potential of transformer-based neural networks in environmental applications.
The findings of this thesis have significant implications for both academic research and practical implementation. By demonstrating the efficacy of the ETSformer neural network in IAQ prediction, the work contributes to the advancement of machine learning methodologies in environmental monitoring. The proposed improvement strategies offer actionable insights for stakeholders, including building managers, policymakers, and health professionals, aiming to reduce indoor air pollution and its associated health risks.
Moreover, the integration of advanced neural architectures with real-world IAQ management exemplifies the potential for interdisciplinary solutions to complex environmental challenges. As concerns about indoor air quality continue to rise globally, especially in the context of urbanization and increased time spent indoors, this research provides a timely and impactful contribution to the field.