中文摘要：

寒冷地区河流冬季常因冰塞问题而成灾，因此对冰塞所涉及的相关问题进行预测和预报对于防灾减灾具有重要的意义。稳封期的平衡冰塞厚度无论对于上游水位的升高还是对后期的开河均具有直接的和重要的影响，用分析型或确定型模型描述非常复杂的河冰过程所导致的冰塞厚度有相当的困难，因此冰塞体厚度描述较多采用的是经验公式，且基本为断面平均厚度。基于人工神经网络的非线形映射特征，尝试将其用于实验室中弯槽段的冰塞厚度分布预测，对数值结果的计算表明，在实际工程中可以采用该方法进行模拟预测。

英文摘要：

Severe flooding or ice-related damage can be resulted from ice jams in cold region. It is important to predict or simulate ice jam evolution in a river for mitigation. Though the possibility where ice jams occur and water level rises in the upstream of river associated with ice jams has been studied recently, equilibrium thickness of ice jam has a decisive influence not only on the rise of water level, but also on the ice breakup, it is difficulty to describe the complex ice-related process with analytical or deterministic models. Normally, simple empirical models may be used to express the average ice jams thickness of cross-section. This paper provides a method to determine the numerical ice jams thickness and its figure in a bend reach based on the nonlinear mapped character of artificial neural network, numerical computation indicates that this the ice jams thickness in bend reach.