|
The weather in Taiwan is high precipitation, and the heavy rainfall is generally in the typhoon season. Therefore, how to forecast stream flow is a worthy research subject for hydraulic engineers to avoid damages by typhoon. In this study, a flow-trend index based on the fuzzy set theory is developed for neural network models to systematically establish the relationship between the rainfall history and stream flow. The precipitation records of Wu-Tu including 18 typhoon events from 1973 to 2001 are used to train and test the stream flow forecasting model. The computing results indicate that the neural network using the flow-trend index is effective in the stream flow predictions.
|