FLOOD FORECASTING METHOD USING RECURRENT NEURAL NETWORK

摘要

The present invention performs machine learning through a circulatory neural network, which is a kind of artificial neural network, by using hydrologic information composed of time series values of rainfall and water level as input information, and estimates the water level at a specific point on the river channel or at the edge of the stream. It is designed to predict whether a point will be flooded or not. Through the present invention, it is possible to accurately and quickly predict the water level and flooding of the predicted point according to rainfall and river level fluctuations, thereby reducing human and material damage caused by flooding.