Enhanced Calibration Techniques for Low-Cost Particulate Matter Monitors

This study advances the field of low-cost air quality sensor calibration by integrating sophisticated machine learning techniques with time series analysis. We combine advanced algorithms such as Support Vector Regression (SVR) and Multilayer Perceptrons (MLP) with SARIMA models to improve calibration accuracy across various sensor types and particulate matter sizes. Our Comprehensive approach reveals the superiority if ensemble methods incorporating SA RIMA components and highlights the varying performance of algorithms across different sensor types and particulate matter sizes. Key findings include the critical importance of temporal features and the persistent challenge in PM<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">10</inf> calibration compared to PM<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2.5</inf>. The results underscore the potential of hybrid approaches that leverage both machine learning and time series analysis, enhancing calibration accuracy and model generalizability across diverse environmental conditions. This work contributes to the development of more reliable and accessible air quality monitoring systems, with significant implications for public health and environmental management.