论文部分内容阅读
提出了用门限回归模型（ＴＲ）来描述年径流时间上复杂的统计特性，并研制了ＴＲ建模的一套简便通用的方案．用作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优化工作这一难题，为ＴＲ模型的广泛应用提供了有力工具．实例的计算结果说明了其可行性和有效性．该方案在其它自然资源中长期预测中也具有广泛的实用价值．
The threshold regression model (TR) is proposed to describe the complicated statistical characteristics of annual runoff time and a simple and universal scheme of TR modeling is developed. Using the improved genetic algorithm, the author can optimize the threshold value and the regression coefficient at the same time, thus solving the problem of a large number of optimization work involved in the TR modeling process and providing a powerful tool for the wide application of the TR model. The calculation result of the example shows its feasibility and effectiveness. The program also has extensive practical value in the medium and long-term prediction of other natural resources.