Classification of Air Pollution Levels using Artificial Neural Network

Abstract

Air pollution can be a threat to the human environment. It becomes a global issue in the world for every country. Air pollution is caused by many factors and becomes dangerous if the concentration level exceeds the normal levels. Several gasses including PM10, SO2, CO, O3, and NO2 can be hazard pollution. These gasses concentration can be sensed by IoT sensors. When the concentration is exceeds the threshold, it become unhealthy condition for human life. This paper proposes to classify air pollution level from IoT data for understanding current condition of air quality. This research proposes neural network methods to classify data into three air pollution levels. The neural network architecture is built from a combination of hidden layers, number of neurons and number of epochs. Based on the experiment, the accuracy of the neural network model can achieve up to 96.61%.