由于不同洪水所估计的马斯京根模型的参数值各不相同,采用相同的参数值势必对洪水演算的精度产生较大的影响。本文提出了一种用于洪水演算的参数自适应马斯京根法,采用数据挖掘中的时间序列相似性搜索方法搜索与当前洪水最相似的洪水过程,利用该相似洪水进行模型参数的最优估计,然后根据所估计的参数进行洪水演算。同时,针对参数估计中所存在的求解复杂、精度差等问题,提出了一种基于免疫克隆选择的马斯京根模型参数估计算法。该方法具有求解精度高,计算速度快、适应性强等特点,能够很好的解决马斯京根模型的参数最优估计问题。仿真实验表明参数自适应马斯京根法具有更高的演算精度,应用实例验证了该方法可行性和有效性。
The estimated parameter values of Muskingum model for the same river reach are typically different in different flood conditions,and statistical analysis also shows that these parameters are highly variable.As a result,if these parameters are specified as constants it is impossible to achieve a high accuracy of flood routing.This paper proposes a parameter adaptive Muskingum method for river flood routing.This method searches the flood sequences most similar to the current flood sequences by using the time-series similarity searching algorithm of data mining filed,then estimates the optimal parameter values of the Muskingum model based on the data of similar flood sequences.The set of parameters so obtained are the best for the current Muskingum flood routing.Aiming at the nonlinearity and complexity of Muskingum flood routing model,this paper puts forth a novel method for the parameter estimation with an immune clonal selection algorithm(ICSA).The simulation and application results show a faster convergence and higher accuracy of ICSA than other techniques such as ant colony algorithm and genetic algorithm,as well as a higher accuracy of the parameter adaptive Muskingum routing than traditional one.