CAJ | 학술논문

针对月降水量高度非线性的特点,以合肥20年的月降水量为时间序列,综合运用径向基函数( RBF)神经网络,建立了一种基于径向基函数的神经网络预测模型。首先对RBF神经网络进行介绍,并将该网络应用于月降水量预测,应用归一化方法对原始数据进行预处理；然后运用MATLAB R2008神经网络工具箱函数建立月降水量预测模型；最后进行仿真实验与分析,将RBF神经网络与传统的BP网络训练预测结果进行比较。结果显示,RBF神经网络模型训练的迭代次数和训练时间、预测结果明显好于传统BP神经网络。
침대월강수량고도비선성적특점,이합비20년적월강수량위시간서렬,종합운용경향기함수( RBF)신경망락,건립료일충기우경향기함수적신경망락예측모형。수선대RBF신경망락진행개소,병장해망락응용우월강수량예측,응용귀일화방법대원시수거진행예처리；연후운용MATLAB R2008신경망락공구상함수건립월강수량예측모형；최후진행방진실험여분석,장RBF신경망락여전통적BP망락훈련예측결과진행비교。결과현시,RBF신경망락모형훈련적질대차수화훈련시간、예측결과명현호우전통BP신경망락。
Owing to the strong nonlinearity of monthly rainfall,taking 1990~2010 monthly rainfall data in the Hefei area as the time se-ries and using the RBF neural network,a new monthly forecast model is developed based on RBF neural network. Firstly,introduce the structure of RBF neural networks and discuss the RBF neural networks application for predicting the monthly rainfall. And then,the func-tions of MATLAB toolbox are adopted to create a network model for the monthly rainfall. Finally,RBF neural network and traditional BP network are compared in their prediction results each other through simulation experiments and studies. Simulation results show that the RBF neural network model is superior to traditional BP neural network.