Author(s): B. Bhattacharya; R. K. Price; D. P. Solomatine
Linked Author(s):
Keywords: Sediment transport; Artificial neural networks; ANN; Model trees; Engelund and hansen
Abstract: A number of sediment transport models have been developed over time. The predictive accuracy of these models, however, is often poor. The transport mechanism is complex, and expressing it through a deterministic mathematical framework is difficult, if not impossible. As an alternative, in this research a data-driven modelling approach to modelling sediment transport is followed. Artificial neural networks and model trees have been used to improve the accuracy of the model of Engelund and Hansen. The input variables to the data-driven models are selected based on the findings of Engelund and Hansen. The models built are tested with a large dataset and their predictive accuracy is found to be better than that of the model of Engelund and Hansen. The data-driven models provide an improved relationship than the original model. A conclusion is reached that the data-driven modelling approach can be suitable for modelling sediment transport.
Year: 2005