Please use this identifier to cite or link to this item: http://hdl.handle.net/123456789/8326
Title: IDENTIFICATION OF RIVER BED PATTERNS USING NEURAL NETWORKS
Authors: Kanth, Katikithala Rajani
Keywords: CIVIL ENGINEERING
RIVER BED PATTERNS
NEURAL NETWORKS
ARTIFICIAL NEURAL NETWORKS
Issue Date: 1998
Abstract: As, the sediment characteristics, the flow characteristics and/or the fluid characteristic are changed in alluvial channels, the nature of the bed surface and water surface changes accordingly. These types of bed and water surfaces are classified according to their characteristics and are called regimes of flow. Regimes of flow has a great influence on such factors as resistance to flow and rate of sediment transport. Based primarily on flume data and some data from natural streams, previous researchers have proposed a few criteria for predidtion of various regimes of flow. However, use of some of these criteria in predicting regimes for recently collected stream data, has demonstrated their limited reliability under varying conditions. Considering the promising results of Artificial Neural Networks (ANNs) in other related fields, a new criterion has been developed to predict various flow classes. ANN attempts to approximate a transfer function that transforms a bounded input vector into a bounded output vector. In the present case, feedforward networks with error back propagation learning rule have been used for training. • Several sets of flume data and field data collected under a wide range of conditions were used. The results of the network are also compared with some of the available regime classfiers. The ANN results are encouraging as the network also performed well on untrained testing data.
URI: http://hdl.handle.net/123456789/8326
Other Identifiers: M.Tech
Appears in Collections:MASTERS' DISSERTATIONS (Civil Engg)

Files in This Item:
File Description SizeFormat 
CED 248028.pdf4.5 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.