Forecast flows in a section of the Bogotá river by artificial intelligent systems

This article presents a comparison between two types of intelligent models: Artificial Neural Networks ANN and Adaptative Neuro-Fuzzy Interference System ANFIS, for forecasting flows in a section of Bogotá (Colombia) river, looking for the most efficient. The simulation was performed in the Matlab computer software, with data collected by hydrological stations of the Corporación Autónoma Regional of Cundinamarca (CAR), from September 2009 to October 2013. The findings suggest that by using artificial intelligence models you can reach a successful outcome, with Correlation Coefficients above 90 % (CC), Mean Absolute Percentage Error (MAPE) below 12 %, Concordance Correlation Coefficient to 84 %, six other statistical evaluating precision and accuracy, suggesting that forecasts will be labeled as good and could think of the use of these techniques in Colombia. Key words—Artificial Neural Networks, Neuro-fuzzy Interference System ANFIS, flood forecasting, flow rate, volume of flood.