年径流预测的遗传门限自回归模型

Genetic Threshold Auto-Regressive Model for Predicting Annual Ruu-Off

为有效利用年径流时间序列资料所隐含的时序分段相依性这一重要信息 ,提出了用门限自回归模型 (TAR)来预测年径流 ,并研制了 TAR建模的一整套简便通用的方案。用所提出的改进遗传算法 ,可同时优化门限值和自回归系数 ,从而解决了 TAR建模过程所涉及的大量复杂寻优工作这一难题 ,为 TAR模型的广泛应用提供了强有力的工具。实例计算的结果说明这套方案是可行的和有效的 ;通过门限值的控制作用 ,TAR模型可以有效地限制模型误差 ,从而保证 TAR模型预测性能的稳健性 ,提高预测精度。该方案具有通用性 。

To effectively utilize the important information of the section interdependence during the time series of annual run off,threshold auto regressive(TAR) model is suggested to predict annual runoff.A simple and general scheme is presented for establishing TAR model.With the improved genetic algorithm by the authors,both of threshold values and auto-regressive coefficients can be optimized ,and the difficulty problem of modeling of TAR is resolved,which gives a strong tool for widely applying TAR model.The case study shows that the scheme is practical and efficient,and that TAR model can successfully reduce model errors,and ensure good stability and accuracy of the model forecasting by controlling threshold valves.As a general method,the scheme has major theoretic valve and wide-ranging application for predicting of nonlinear time series.