预测低温冷害的门限回归模型

Threshold regressive model for forecasting low temperature disaster

提出用门限回归模型 (TR)来描述低温灾害指标变量在灾害状况与非灾害状况的突变性。研制了 TR建模的一套简便通用的方案 ,用笔者改进的遗传算法可同时优化门限值和回归系数 ,从而解决了 TR建模过程所涉及的大量复杂寻优工作这一难题 ,为 TR模型的广泛应用提供了有效工具。实例的计算结果说明了其可行性和有效性。该方案具有通用性 。

Threshold regressive (TR) model is recommended to describe the mutation of low temperature index variable from normal situation to disaster situation, and a simple and general scheme is presented to deduce TR model Both threshold values and regressive coefficients of TR model can be optimized with the genetic algorithm improved by the authors, and the difficulty problem of modeling of TR can be resolved, which gives a effective tool for widely applying TR model The example shows that the scheme is practical and efficient As a general method, the scheme can be applied to forecasting time series of various natural disaster