Data mining approach for friction factor in mobile bed channel

Abstract

Resistance to flow in mobile bed channels varies between wide limits because the form of the boundary roughness, as well as resistance to flow and sediment transport, is a function of the fluid, flow, bed material and channel characteristics. A system can be evaluated analytically if the relationships in the model are simple enough to be represented in a mathematical form. Complex systems such as mobile bed friction factor can be investigated through the data mining technique. Data mining is currently utilised in almost all branches of science as an alternative and complementary model to traditional physically based modelling systems. This paper proposes a genetic algorithm optimised back-propagation neural network to model the friction factor. Based on the weights of the neural network an attempt is made to quantify the contributions of the different parameters in the friction factor prediction.