一种改进的不确定性水质模型参数率定方法

摘要

为了能够使得水质模型的参数率定更符合实际情况,减少水质模型当中常出现的异参同效的现象,提出一种能够结合实验与算法共同对模型参数进行识别的方法,即通过设置约束条件,将单纯的"与实测值进行对比"的一般率定过程转化为"能够对模型内部过程进行一定程度控制"的率定过程.以WASP模型在北运河的某河段中进行的水质模拟结果表明,通过引入约束条件可以从更为系统的角度有效地去除部分不符合污染物转化规律的参数组合,使得最终水质模型能够更为准确地反映实际的水质变化过程,同时也减少了模型的不确定性与模型参数率定的"异参同效"现象;在模型的非线性结构与约束条件的作用下,相同子模块中原本独立的参数会呈现一定的相关性;随着对区域的水质变化过程研究的深入,当引入新增约束条件或是降低容错比例时,约束条件的作用将会进一步增强.%In order to make the parameters of the water quality model more realistic and reduce the equivalent phenomenon of different parameters in water quality model, this paper proposed a method to identify the model parameters by setting laboratory experimental constraints, which could be used to make a certain degree of internal process control. The example of a water quality simulation by WASP model in a section of the North Canal showed that by this method, the model could be more accurate to reflect the actual water quality change process by reducing uncertainty and equivalent from combination of the parameters. Through the action of the nonlinear model structure and constraint conditions, the originally independent parameters under the same sub module began to demonstrate certain correlation. With the further study of the water quality change process, the constraint conditions would be further enhanced when introduced new constraint conditions or fault decrease tolerance.