Machine Learning Based Air Pollution Monitoring And Forecasting with IOT System

In recent times air pollution has emerged as a major health hazard in Dhaka city. To understand and mitigate this major problem requires a multi-step approach. The first step is to determine the exact nature of the air particulates that are responsible for the air pollution health hazard in Dhaka city. To this end we have built an air pollution monitoring IOT device that collects data on the composition of the different air pollution particulates in Dhaka city. The second step is to determine the source for the air pollution particulates. Using our collected data along with existing CASE (Clean Air Sustainable Environment)’s data and dataset provided by Bangladesh Meteorological Department(BMD), we calculated the air pollutant back trajectories using NOAA’s Hysplit computer model. The third step is to build an air pollution prediction model using the Hysplit computed air pollution back trajectories and the air pollution monitoring data. We developed a machine learning based classifier that predicts future air pollutant concentration, PM2.5 and other air pollutants listed in the CASE dataset, based on air stream trajectory and meteorological data. This air pollution prediction capability will aid in developing policy to mitigate the air pollution health hazard in Dhaka city.
Department Name
Electrical and Computer Engineering
North South University
Printed Thesis