摘要
以乌尔逊河为例,采用相空间重构理论计算实测月降雨的延迟时间、嵌入维数、G-P饱和关联维数和Laypunov指数,证明月降雨时间序列存在混沌现象。运用LS-SVM模型对乌尔逊河月降雨混沌时间序列进行预测,并利用交叉验证法求取LS-SVM模型两个重要参数的最佳组合,同时与RBF神经网络模型进行了对比分析。结果表明,在做混沌时间序列分析时LS-SVM模型的预测精度优于RBF神经网络模型。
Taking Wuerxun River for an example, phase space reconstruction theory is adopted to calculate the delay time, embedded dimension, G-P saturated correlation dimension and Laypunov exponent of the observed monthly rain- fall. And it proves that the monthly rainfall time series exist chaotic phenomena. Then the model of least squares support vector machine （LS SVM） is used to predict the monthly rainfall chaotic time series of Wuerxun River and the cross validation method is applied to obtain optimal combination of two important parameters. Compared with radial basis function （RBF） neural network model, the results show that the prediction accuracy of LS-SVM is higher than that of RBF for prediction of chaotic time series.
出处
《水电能源科学》
北大核心
2012年第9期13-16,214,共5页
Water Resources and Power
基金
国家自然科学基金资助项目(50569005
40901262)
国家水体污染控制与治理科技重大专项基金资助项目(2009ZX07106-006)
