Enhancement of a Neuro-Fuzzy Models Using Ant Colony Optimization for the Prediction Level of CO Pollution

Air pollution has a non-linear behaviour over a day, in which can be detected patterns of behaviour to shape them, a way of doing this is using fuzzy logic, as is in this case, where the membership functions are generated based on groups performed by the algorithm fuzzy c-means. These groups will be the knowledge base for the membership functions of the fuzzy system trained by Anfis, which is a Neuro-Fuzzy system and allows us to have greater approximation in the prediction, however, it is proposed to improve the results using the ant colony optimization algorithm, three models from Mexico city pollution are developed, and by the probabilistic properties provided by the pheromone evaporation in the ACO algorithm, the error level is reduced, therefore, a better prediction is made.