Environmental air pollution monitoring in non industrial area using machine learning techniques and IOT

Abstract

Objectives: To provide an enhanced embedded-based IOT network for monitoring environmental pollution in non-Industrial areas with an efficient machine learning pollution prediction system. Methods: The methodology of the Dual processing Environmental Monitoring System (DPEMS) is carried out through a Dual processing unit (Arduino-Raspberry Pi) with advanced environmental air pollution, collecting sensors such as DHT22, CO2 (MG811), NO2 (MICS-4514), and SO2 (SGS-SO2). The environmental air pollutant data has been shared with IOT cloud storage from the Dual central processing unit to the IBM blue mix platform. To enhance a better pollution prediction system, machine learning classifiers such as ANN, SVM, and Decision Tree has been applied. The machine learning training and testing validation has been done using Pycharm 2021.1.1. The actual and predicted pollutant value has been evaluated using the performance metrics as RMSE, R2, and IA. Findings: the proposed IOT-based embedded DPEMS is utilized to increase the accuracy of real-time actual pollutant value and alert the threshold level of pollutant particles such as Temperature, Humidity, Carbon dioxide (CO2), and Nitrogen dioxide (NO2), and sulfur dioxide (SO2) in Non-Industrial areas.