Analytical equations based prediction approach for PM2.5 using artificial neural network

The worldwide, particulate matter pollution is considered one of the deadliest types of air pollution due to its significant impact on the global environment and human health. The particulate matter (PM2.5) plays a key role in evaluating the air quality index. However, the conventional PM2.5 monitoring instruments used by the air quality monitoring stations are costly, bulkier, time-consuming, and power-hungry. Furthermore, due to limited data availability and non-scalability, it is challenging to provide high spatial and temporal resolution in real-time. To overcome these challenges, we present analytical equations based prediction approach for PM2.5 using an artificial neural network. Moreover, we contribute the correlation study between PM2.5 and other pollutants using a large authenticate data set of Central Pollution Control Board online station, India. The correlation study reveals the strong correlation of eight pollutants with PM2.5, which found useful for the proposed prediction model and future research work. The computation of the proposed analytical equation using a low-cost processing tool (excel sheet) demonstrates a good match between predicted and actual results. Additionally, the derived analytical equation for the prediction can be computed using a wireless sensor node which ultimately eliminates the need for costly propriety tools. The performance of proposed analytical equation for prediction show root mean square error and coefficient of determination (R2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$R^2$$\end{document}) 1.80 μ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document}g/m3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^3$$\end{document} and 0.99 respectively using eight correlated predictors. The recalibrated prediction model with three correlated predictors show RMSE of 7.54 μ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document}g/m3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^3$$\end{document} and R2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$R^2$$\end{document} of 0.97 and proves the effectiveness of the proposed approach in implementation using minimum power-hungry gas sensors on the WSN. Therefore, obtained results demonstrate that the proposed approach is one of the promising approaches for monitoring PM2.5 without power-hungry gas sensors and bulkier analyzers.