In order to forecast and classify environmental risks, artificial intelligence (AI) techniques were applied to the air quality problem. Predetermined gaseous pollutant
concentration data were acquired with the intent of predicting the risks. Such concentrations are denominated air quality indicators, and are regulated all around the world, including by brazilian law. The data concerning these indicators were used in a model that consists of two
AI techniques: artificial neural networks and particle swarm optimization. The air quality indicators concentration prediction resulted in one day ahead values. The risk modeling utilizes the predictions as inputs values, correlating them in order to obtain the resulting air quality and, the risk that such quality has upon the human health. The risk model is based on a third AI technique, called fuzzy logic. The present work obtained two main results. The first was the accurate forecasts made by the prediction model. The second was the achievement of a coherent classification of the risks.
