科研成果详情

Chaos Theory describes systems that are extremely sensitive to initial conditions. Assuming that weather is a chaotic phenomenon, chaos theory may provide suitable weather prediction models. The study describes chaos theory-based weather prediction model for forecasting short range severe rainstorms in Hong Kong. The model uses a Lee oscillator, which is based on chaos theory, as a transfer function in a Multi-Layer Perceptron model. The proposed model was compared with three non-linear time series models, RBF, SVM and MLP, and was more accurate by 4.518%, 4.902% and 3.791% respectively.