Seite - 907 - in Book of Full Papers - Symposium Hydro Engineering

Fig. 1 Process of PCA-RBF neural network 2.2. RBF NEURAL NETWORK RBF neural network is a kind of three-layer feedforward neural network, including input layer, hidden layer and output layer. The network uses radial basis function as the "base" of the hidden element. By this method, the input vector can be directly mapped to the hidden layer. The transformation from the input layer to the hidden layer is nonlinear mapping, and the transformation from the hidden layer to the input layer is linear mapping. The network is nonlinearly mapped from the input layer to the output layer, but the output layer is linear to the adjustable parameters. Thus, the weights of the network can be directly solved from the linear equations, which can effectively improve the learning efficiency of the network and avoid falling into the local minimum. Generally, Gaussian function is chosen as the activation function of the hidden layer, and the output of the each hidden layer is:   2 2exp 2ij j i ja c x     ， 1,2 ,,j N  [1] 907