Prediction of Daily Concentration of Carbon Monoxide based on PCA-ANN

Abstract

Air pollution crisis is one of the gifts of industrial and technological development. It is increasing with population growth, urbanity, and consumption of fossil fuels. Air pollution crisis put the people who inhabit in polluted areas in danger. Although senior government managers have already done a lot of affairs to inhibit production and decrease the risks of all kinds of pollutants, they should survey the research issues relating to the solution of air pollution crisis at first to control and anticipate the amount of pollutant indexes. In this paper, the effects of PCA (principle component analysis) on the efficiency of two kinds of Artificial Neural Network (MLP, ANFIS) are studied to anticipate the daily density of CO in the weather of Tehran city. This research shows that the decrease in many variables and the recognition of effective variables on ANN improve performance neural network training speed.