地下水位动态预测的遗传门限回归模型

Genetic Threshold Regressive Model for the Dynamic Predication of Groundwater Level

提出了用基于遗传算法的门限回归模型 (TR)描述地下水位及其影响因子之间复杂非线性关系的新方法。实例计算的结果说明 :通过门限值的控制作用 ,TR模型可以有效地限制模型误差 ,保证了模型预测性能的稳健性 ,提高了预测精度 ;该方法在其它非线性时序预测中具有广泛的实用价值。

The threshold regressive(TR) model is suggested to describe the complex nonlinear relationship between groundwater level and its influence factors.A general and simple scheme is presented for the TR modeling.From the example it is shown that the scheme is practical and efficient,and that the model error can effectively controlled by the TR model and the stability and accuracy of the model forecasting can be ensured with the threshold values.The said method can be used to forecast the nonlinear time series in practice.