The flood forecasting is the key to support the right decision making. A method to forecast flood accurately and timely are important. We propose a method based on Radial Basis Function (RBF) neural network which has the important application in flood water level forecasting. The traditional way of training of the neural network may drive the network to converge in local minima instead of global minimum. We introduce a cuckoo search algorithm to train parameters of neural network instead of a normal way. We implement our proposed algorithm where the input is the real data from Little Wabash River. In the experimental part, we choose the type of Radial Basis Function to be Gaussian and Polyharmonic. We investigate the impact of these two RBF functions and discuss the error between the forecast and the actual values. We conclude that Polyharmonic function suits to this problem better than Gaussian function.
展开▼
