基于混沌支持向量机的径流时间序列预测研究——以渭河宝鸡林家村站径流序列为例

混沌理论和支持向量机具有强大的非线性处理能力.首先利用混沌系统相空间延迟坐标重构理论对林家村站月径流进行相空间重构,以便更为深刻地挖掘月径流序列中的信息,并运用最大Lyapunov指数法证实渭河林家村站月径流系列具有混沌特性.在此基础上利用基于统计学习理论的支持向量机建立混沌时间序列的预测模型.仿真结果表明,所提出的模型预测结果好于混沌神经网络模型的预测结果,该模型具有较高的泛化能力,可用于林家村站月径流预测.

Research on prediction of runoff time series based on SVM——Case study of runoff of Linjiacun gauging station on the Weihe river at Baoji city

Chaos theory and support vector machine have great nonlinear treatment capability.Firstly the phase-space reconstitution of monthly runoff at Linjiacun gauging station was made by phase-space delay coordinate reconstitution theory of chaos system,so that information of monthly runoff series was profoundly investigated.At the same time,the chaotic feature of the monthly runoff at Linjiacun is proved by using maximal Lyapunov exponent method.With this as the basis,the prediction model of chaos time series is built by using support vector machine from statistic learning theory.It is shown by simulation experiments that the prediction results of the model proposed are better than the ones of Chaos ANN,and this model has better generalization,and may be used for monthly runoff prediction at Linjiacun.