月径流序列混沌特性识别及预测

Chaotic characteristics identification and prediction of monthly runoff series

以混沌理论和相空间重构原理为基础,分析计算大峪水文站1955-2006年月径流序列的最佳延迟时间和嵌入维数;运用最大Lyapunov指数λ1>0证实大峪月径流序列具有混沌特性,从而建立了基于混沌特性的BP神经网络预测模型。仿真及预测结果表明:该模型预测精度较高,可用于大峪月径流预测。

Based on chaos theory and phase-space reconstitution principle,the best delay time and embedding dimension of monthly runoff series were analyzed and calculated from 1955 to 2006 at DaYu hydrological station;The maximum Lyapunov index λ10 was used to confirm that the monthly runoff series has chaotic characteristics,thus,BP neural network prediction model based on chaotic characteristics was built.The result of simulation and prediction showed that the predicted precision of this model is better and can be used for monthly runoff prediction at DaYu.