Self-Organizing Polynomial Neural Network for Modelling Complex Hydrological Processes (No. R861)
Access status:
Open Access
Type
Report, ResearchAbstract
Artificial neural networks (ANNs) have been used increasingly for modelling complex hydrological processes. In this paper, we present a self-organizing polynomial neural network (SOPNN) algorithm, which combines the theory of bio-cybernetic self-organizing polynomial (SOP) with the ...
See moreArtificial neural networks (ANNs) have been used increasingly for modelling complex hydrological processes. In this paper, we present a self-organizing polynomial neural network (SOPNN) algorithm, which combines the theory of bio-cybernetic self-organizing polynomial (SOP) with the artificial neural network (ANN) approach. With the SOP feature of seeking the best combination of polynomial model parame-ters through optimal reduction of the partial polynomial nodes in the network and the ANN function of seeking minimum error at each layer of the network, the algorithm possesses superiority in nonlinear modelling of dynamic systems. The developed al-gorithm is applied to model a real-time rainfall-runoff process. The results demon-strate the capability of the SOPNN approach in addressing difficult issues of ANN modelling: selection of appropriate model inputs, optimization of the network struc-ture and error minimization. The comparison of modelling results shows that the SOPNN algorithm performs better in complex hydrological modelling than two other nonlinear approaches: the group method of data handling (GMDH) algorithm and the traditional back-propagation network (BPN) algorithm.
See less
See moreArtificial neural networks (ANNs) have been used increasingly for modelling complex hydrological processes. In this paper, we present a self-organizing polynomial neural network (SOPNN) algorithm, which combines the theory of bio-cybernetic self-organizing polynomial (SOP) with the artificial neural network (ANN) approach. With the SOP feature of seeking the best combination of polynomial model parame-ters through optimal reduction of the partial polynomial nodes in the network and the ANN function of seeking minimum error at each layer of the network, the algorithm possesses superiority in nonlinear modelling of dynamic systems. The developed al-gorithm is applied to model a real-time rainfall-runoff process. The results demon-strate the capability of the SOPNN approach in addressing difficult issues of ANN modelling: selection of appropriate model inputs, optimization of the network struc-ture and error minimization. The comparison of modelling results shows that the SOPNN algorithm performs better in complex hydrological modelling than two other nonlinear approaches: the group method of data handling (GMDH) algorithm and the traditional back-propagation network (BPN) algorithm.
See less
Date
2005Publisher
School of Civil Engineering, The University of SydneyLicence
Copyright All Rights ReservedRights statement
This publication may be redistributed freely in its entirety and in its original form without the consent of the copyright owner. Use of material contained in this publication in any other published works must be appropriately referenced, and, if necessary, permission sought from the author.Faculty/School
Faculty of Engineering, School of Civil EngineeringDepartment, Discipline or Centre
Centre for Advanced Structural EngineeringShare