Forecasting Air Pollution by Adaptive Neuro Fuzzy Inference System

Air pollution causes a variety of adverse effects on humans such as illness or even death and damages the living organisms and the natural environment. This environmental issue needs to be controlled using various application and technology to estimate the composition of multiple pollutants in the atmosphere for a specified time and location. The present study aims to develop a system for air pollution forecasting using an adaptive neuro-fuzzy inference system. This method is a type of artificial neural network that integrates both neural networks and fuzzy logic principles. The adaptive neuro-fuzzy inference system calculations include four phases including implement fuzzy system, enter parameters, start the learning process, and verify the processed data. As a sample, the concentrations of atmospheric pollutant data recorded by sensors. The adaptive neuro-fuzzy inference system method predicts four air pollution indicator levels, including carbon monoxide, sulfur dioxide, nitrogen oxides, and trioxygen. The analysis results reveal that the mean absolute error of the adaptive neuro-fuzzy inference system method results is less than 15 %.