基于混合递阶遗传算法的径向基神经网络学习算法及其应用

RBFNN algorithm based on hybrid hierarchy genetic algorithm and its application

在研究径向基神经网络学习算法的基础上 ,提出了一种新型的径向基神经网络学习算法———混合递阶遗传算法 .该算法将递阶遗传算法和最小二乘法的优点结合在一起 ,能够同时确定径向基神经网络的结构和参数 ,并具有较高的学习效率 .采用基于混合递阶遗传算法的径向基神经网络对混沌时间序列学习和预测 ,取得了较好的效果 .

Based on the study of RBFNN (radial basis function neural network) training algorithm and genetic algorithm, a new RBFNN training algorithm-hybrid hierarchy genetic algorithm is introduced by combining hierarchy genetic algorithm and least_square method. The hybrid algorithm greatly increases the training speed while is still able to determine the structure and parameters of the RBFNN from sample data. The new training algorithm is used to identify and predict M_G chaos time series, and the simulation gives staisfied result.