| 摘要: |
| 讨论研究了水文特征值预报的数学方法:统计回归模型、神经网络模型和模糊回归模型.三个计算实例表明如果系统的线性关系较好,统计回归模型的结果最好;如果系统的线性关系差,神经网络模型的结果最好;如果用于率定模型的资料太短.任何一个模型都不可靠. |
| 关键词: 洪水预报 神经网络方法 回归分析 模糊回归 |
| DOI：10.18307/1997.0204 |
| 分类号: |
| 基金项目: |
|
| Comparison on Three Mathematical Models For Special Values in Flood Forecasting |
|
Li Zhijia1, Kong Xiangguang2
|
|
1.Department of Water Resoures & Hydrology, Hohai University, Nanjing 210024;2.Yishusi Civil Engineering Bureau, Ministry of Water Resources, Xuzhou 221009
|
| Abstract: |
| Three mathematical models, i. e. regressive analysts method, artificial neutral method and fuzzy regressive method, are commonly used in the flood forecasting for special values. The practical calculation results of three cases provided in this paper show that either one is suitable for all cases. The regressive analysis method is favorable when the system has better linear correlations; otherwise the artificial neutral net method is better if the system is not linearly correlated. None of the above-mentioned method is reliable when the data needed for the calibration are not enough. |
| Key words: Flood forecasting artificial neutral method regressive analysis fuzzy regressive analysis |
