克鲁伦河月径流混沌时间序列的LS-SVM和RBF预测

采用相空间重构理论计算实测月径流的延迟时间、嵌入维数、G-P饱和关联维数和Laypunov指数,证明克鲁伦河月径流时间序列存在混沌现象。混沌时间序列的最小二乘支持向量机预测模型涉及参数较少,计算过程简便,训练速度快。RBF神经网络预测模型具有较快的训练速度和较强的非线性映射能力。同时,二者在建模中都引用了径向基函数,从而更加简化了非线性问题的求解。实例表明:将这两种模型应用在月径流时间序列预测上,其运算速度都很快,但在预测精度上,最小二乘支持向量机预测模型要优于径向基神经网络预测模型。

Prediction of the monthly runoff chaotic time series in Kelulun River using least squares support vector machines and RBF neural network models

Based on the phase space reconstruction,the paper calculated the best delay time,real monthly runoff,embedded dimension,G-P saturated correlation dimension and Lyapunov exponent,and proved that the monthly runoff time series of Kelulun River chaos phenomena exist.Prediction model of Least squares support vector machine of chaos time series,it is involving less parameters,is simple of calculation process and has high speed training.Prediction model of RBF neural network has faster training speed and strong non-linear mapping ability.At the same time,radial basis function has introduced in the course of setting up the models of least squares support vector machine and radial basis function,it has simplified the course of solving the non-linear problems.The instance indicates that the two models are used in monthly runoff time series forecast,its operation speed is very fast,but in the operation accuracy,the prediction model of least squares support vector machine is better than radial basis function.