A fuzzy neural network prediction model based on manifold learning to reduce dimensions for typhoon intensity
Authors:
- 1 June 2013
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 562-566
Abstract
In order to explore the new forecast method for typhoon, using the isometric mapping(ISOMAP) feature extraction algorithm with manifold learning to dimension reduction calculation processing for the high dimensional data of typhoon intensity forecast factor, and the forecast typhoon intensity model input is constructed combined with the screening method of stepwise regression forecast factor, and further the 12-h to 72-h models are researched by using the method of fuzzy neural network(FNN). The 5-10 month prediction of western Pacific (WP) typhoon intensity with 10 years from 2001 to 2010 have been performed in the methodological experiment of nonlinear meteorological FNN prediction modeling in this paper, and 36 typhoon intensity forecast models have been established with a monthly intervals 12-h, and the last 30 independent samples would be forecast by each prediction model for experiments. Further under the circumstance of the same sample cases, the 36 prediction results of FNN model based on ISOMAP and the climatology and persistence (CLIPER) model are compared, and the result is proved that the former is obviously more skillful than the latter.Keywords
This publication has 7 references indexed in Scilit:
- A Nonlinear Artificial Intelligence Ensemble Prediction Model for Typhoon IntensityMonthly Weather Review, 2008
- A New Supervised Manifold Learning MethodJournal of Computer Research and Development, 2007
- On the Predictability of Tropical Cyclone Tracks in the Northwest Pacific BasinMonthly Weather Review, 2003
- Statistical Prediction of Tropical Cyclone Motion: An Analog–CLIPER ApproachWeather and Forecasting, 2002
- Recent progress on tropical cyclone research in ChinaAdvances in Atmospheric Sciences, 2002
- A Global Geometric Framework for Nonlinear Dimensionality ReductionScience, 2000
- Approximation accuracy analysis of fuzzy systems as function approximatorsIEEE Transactions on Fuzzy Systems, 1996