<fmt:message key='jsp.layout.header-default.alt'/>  

DSpace@UM >
Faculty of Computer Science and Information Technology >
Conference Proceedings >
International Conference on Informatics >
Informatics 2007 >

Please use this identifier to cite or link to this item: http://hdl.handle.net/1812/340

Authors: Ramani Bai. V
Chaw Seng, Woo
Faridha, Othman
Gopinath Ramadas
Keywords: Optimization
Reservoir operation
Multi-layer feed forward network
Data mining
Training set
Flood Control
Issue Date: 2007
Abstract: Artificial neural networks have strong data fitting capability. In domains where explaining rules are critical, such as release of water from dam, denying loan applications etc., classical neural networks are not the tools of choice. ‘Neural Networks Cannot Explain Results’. This is the biggest criticism directed at neural networks or a challenge directed at using neural networks in water resources engineering. The main goal of this research is through processing of data (records from the past) to describe the underlying dynamics of the complex systems and predict its future. One of the solutions is data mining that is sorting through data to identify patterns and establish relationships. Using the best represented data from several previous time steps, a more complex data-driven model on artificial neural network can be built. A problem related to operation of water reservoir is selected to provide a better data representation to the network to evolve better results and continuity of the system performance compared to earlier works. This paper illustrates an artificial developmental system that is a computationally efficient technique for the automatic generation of complex Artificial Neural Network (ANN), which can better represent all the features of the data. MLFFNN models of operation of dam are developed by representing the training set to the network in following four different forms viz. 1)Historical data without optimization given as monthly (HANN), 2)Classical method of optimized results given as monthly (ANN), 3)Implicit method with optimized results given as yearly (IANN), and 4) Explicit method with optimized results given as monthly (EANN). Results have shown that neural network estimates are sensitive to sample representation, but are robust in terms of network architecture. Also the comparison to conventional statistical models, show the superiority of this approach of using ANN. In addition, this research offers an effective and reliable approach that can point out the best direction for maintaining continuity course of operation and hence with significant benefit to the decision makers on water release from the dam. The presented approach to model approximation may be used in various schemes of water resources optimization.
Description: Proceeding of the 2nd International Conference on Informatics (Informatics 2007), 27th-28th November 2007, Hilton Petaling Jaya Hotel, Petaling Jaya, Selangor, Malaysia. T1-14 - T1-20
URI: http://dspace.fsktm.um.edu.my/handle/1812/340
Appears in Collections:Informatics 2007

Files in This Item:

File Description SizeFormat
AIA.pdf3.42 MBAdobe PDFView/Open

This item is protected by original copyright

Your Tags:


  © Copyright 2008 DSpace Faculty of Computer Science and Information Technology, University of Malaya . All Rights Reserved.
DSpace@UM is powered by MIT - Hawlett-Packard. More information and software credits. Feedback