Radial basis function network for prediction of hydrological time series

A. W. JAYAWARDENA, Pengcheng XU, Wai Keung LI

Research output: Chapter in Book/Report/Conference proceedingChapters

1 Citation (Scopus)

Abstract

In this study, a network using radial basis functions as the mapping function in the evolutionary equation for prediction of time series is presented. A radial basis function network requires the determination of the number of centres of the radial basis functions, their receptive field widths, and the linear weights of the network output layer. Methods to estimate the widths of the receptive fields, and the number of centres for the radial basis functions are introduced in the study. The latter is based on the concept of the Generalized Degrees of Freedom. The linear weights are determined by the least squares method. The predictions by the proposed method when compared with the actual values of four hydrometeorological data sets, are better than those by the traditional approach of fixing the number of centres. Copyright © 2003 International Association of Hydrological Sciences.
Original languageEnglish
Title of host publicationWater resources systems: Water availability and global change
EditorsStewart FRANKS, Gunter BLOSCHL, Michio KUMAGAI, Katumi MUSIAKE, Dan ROSBJERG
Place of PublicationWallingford, Oxfordshire
PublisherInternational Association of Hydrological Sciences
Pages260-266
ISBN (Print)9781901502279
Publication statusPublished - 2003

Citation

Jayawardena, A. W., Xu, P., & Li, W. K. (2003). Radial basis function network for prediction of hydrological time series. In S. Franks, G. Bloschl, M. Kumagai, K. Musiake, & D. Rosbjerg (Eds.), Water resources systems: Water availability and global change (pp. 260-266). Wallingford, Oxfordshire: International Association of Hydrological Sciences.

Keywords

  • Chao Phraya River, chaos
  • Generalized degrees of freedom
  • Mekong River
  • Phase space
  • Radial basis functions
  • S-index

Fingerprint Dive into the research topics of 'Radial basis function network for prediction of hydrological time series'. Together they form a unique fingerprint.