基于Bayes方法和模糊检验的年最高水位的概率模型研究

提出了年实测最高洪水位分布概率模型的模糊检验方法，并以此分布为先验分布，用Bayes方法得到验后分布的统计参数，以进一步修正概率模型的统计参数，得到更新的模型参数．计算实例中，以长江下游某市水文站实测的58年最高洪水位资料为样本，用K-S法和模糊检验法对年最高水位的分布规律进行检验，其结果表明：用K—S法检验，年最高水位概率分布对正态分布和极值I型分布都不拒绝，但通过模糊检验法可以证实正态分布更接近实际；再利用后10年的实测最高洪水位的小样本，对模型的分布参数进行了修正．提出的年最高洪水位的更新概率模型及参数计算方法，为现有堤防工程的加高设计方案决策提供了理论基础．

Study on probability model of annual highest water level based on the Bayes Method and Fuzzy Test

A fuzzy test method for the distribution probability model of measured annual highest flood water level is proposed which the tested distribution is a prior distribution, statistical parameters of the probability model can be updated and renewed by post-check parameters by the Bayes method. For a numerical example, using K-S test method and the fuzzy test method, the highest measured flood water levels in 58 years at a city hydrological station in the downstream of Yangtze River is used for the statistic samples, and the distribution pattern of annual highest water level are tested. It is found that a normal distribution and a type I extreme value distribution are not be refused by the probability distribution of annual highest water level using K-S test method, but it is confirmed that the normal distribution is more approachable actual situation by fuzzy test method. Then, the distribution parameters of the probability model are amended using the small sample of the measured highest flood water level of last 10 years. The proposed calculating method of renewed probability model and parameters for distribution of the annual highest water level in this paper can afford a theory method for heightening design plan decision of existing dikes.