Application of artificial neural network to predict the friction factor of open channel flow

The friction factor of an open channel flow is generally affected by the Reynolds number and the roughness conditions, and can be decided by laboratory or field measurements. During practical applications, researchers often find that a correct choice of the friction factor can be crucial to make a sound prediction of hydraulic problems. In this paper, a three-layer artificial neural network (ANN) was set up to predict the friction factors of an open channel flow, with the Reynolds number and the relative roughness as two input parameters. The Levenberg–Marquardt (LM) learning algorithm was employed to train the model by using laboratory experimental data, and the trained network was tested by a single set separated from the rest of the data and a good correlation between the experimental and predicted results has been obtained. Finally, the ANN simulated results were compared with the calculated results obtained by the empirical formula and both comparisons showed that the ANN model can be used to predict the non-linear relationship between the friction factor and its influencing factors correctly once enough samples are provided. The successful application proved that ANN model can be used in engineering practice as a convenient and effective method, and those traditional hydraulic problems which are mostly based on laboratory tests can be analyzed by ANN modelling.