Application of RBF Neural Network Model on Forecasting Tourists Quantity in Hainan Province - Advanced Research on Computer Education, Simulation and Modeling - page 306

has not good ways now. As the number of visitors has a good nonlinear characteristic
and the RBF neural network is better used to handle nonlinear problems, the RBF
neural network can apply to forecast the number of tourisms. The article establishs
prediction model and predictes the number of tourisms .The result of experiment proves
that the prediction model has good prediction effect.

1 The Principle, Structure and Algorithm of RBF Neural Network

1.1 The Principle of RBF Neural Network

RBF neural network is the abbreviation of radial basis function neural network, which
is a kind of feed-forward neural network. Its construction is based on the function
approximation theory. The distance ||dist|| between weight vector and threshold vector
is used to independent variable of the transfer function of the network “adbas”. The
||dist|| is got through the product of input vector and weighted matrix's row vector. Each
hidden layer neurons transfer function of RBF neural network makes up a base function
of a fitting plane, so RBF neural network gets the name.

1.2 The Structure of RBF Network

RBF radial basis network is a three-layer feed-forward neural network, which includes
an input layer, a hidden layer with radial basis function neurons and an output layer
with linear neurons. As shown in figure 1 [4].
Hidden layer is usually using radial basis function as excitation function and the
radial basis excitation function is commonly gaussian function, which is usually ex-
pressed as:

( )

( )

2

⎡ ⎤

q q

xc w 1 X b 1

R − = exp − − ×

⎦ (1)

⎢ ⎥

i

⎣

i i

is the Euclidean distance, c is the center of gaussian function.

q

Where w 1 X

−

i

Xx

q

= (

q

, x

q

,..., x

q

,..., x

q

)

is the qth input data.

12 j m

input layer hidden layer output layer

Fig. 1. The structure of RBF neural network