黄河内蒙段凌情变化规律及智能耦合预报模型

Variation laws of ice regime and intelligent coupling forecast model in the Inner Mongolia reach of the Yellow River

由于特殊的地理位置、水文气象条件及河道特性的影响,黄河内蒙段几乎每年都会发生凌汛。对黄河内蒙段主要控制站的气象水文等实测数据进行分析后,发现近年来随着凌期气温升高,流量增大,流凌、首封日期推后,开河日期提前,且最大冰厚明显变薄。为此,以黄河内蒙段巴彦高勒站为例,通过相关分析选取合适的预报因子,采用基于遗传算法的神经网络方法建立了凌情智能耦合预报模型(GA-BP模型),对流凌、封河、开河日期进行预报。对比不同模型的预报结果,发现多元线性模型、BP模型和GA-BP模型合格率分别为80%、86.7%和93.3%,GA-BP模型的预报精度较高。因此,GA-BP模型可以为黄河内蒙段的凌汛灾害防治提供重要支持。

Due to the special geographical position,hydro-meteorological conditions,and river course characteristics,ice flood almost occurs every year in the Inner Mongolia reach of the Yellow River.The meteorological and hydrological data at the main controlling stations in the Inner Mongolia reach were analyzed,which suggested that the temperature and flow discharge increase in recent years,the ice run date and freeze-up date push back while the break-up date brings forward,and the maximum ice thickness thins obviously.In this paper,the appropriate prediction factors were selected by the correlation analysis.Ice regime intelligent coupling forecast model was built using the neural network method based on the genetic algorithm.The model was applied to forecast the ice run date,freeze-up date,and break-up date at the Bayangaole station in the Inner Mongolia reach.The forecast results obtained from different models were compared,and it showed that the multiple linear model,BP model,and GA-BP model have high passing percentages,which are 80%,86.7%,and 93.3%,respectively.GA-BP model has the highest forecast accuracy and can provide the critical support for the prevention of ice disasters in the Inner Mongolia reach of the Yellow River.