AI Summary & Key Takeaways
AI InsightIt was shown that ANN estimations for gravel bed load data are more successful than the sand load data.
📌 3 Key Takeaways
Abstract
Equations used for calculating bed load transport rates generally assume steady flow conditions. This assumes that the relationship of bed load transport as a function of water discharge, or some other flow parameter such as flow depth or boundary shear stress, is single-valued. One of the reasons for adopting such an approach is that almost all the pertinent laboratory data on bed load transport have been obtained from experiments performed under steady flow conditions. Similarly, the scarcity of accurate bed load field data obtained during the passage of floods is attributed to the difficulties, which at times can become life threatening, encountered under such conditions. Provision of data under difficult conditions may lead to inability to provide data in some cases and interruption in data continuity. It is extremely difficult to make predictions using classical statistical science in discontinuous or lack of data situations. Artificial neural networks (ANN) is a usefull tool to use in prediction inefficient or data conditions. In this study two Artificial Neural Network (ANN) methods, radial basis functions and generalized regression neural network are employed to estimate the bed load data. It was seen that the ANN estimations are more satisfactory compared to those of the conventional statistical methods results. It was shown that ANN estimations for gravel bed load data are more successful than the sand load data.