用模拟退火算法估计水质模型参数

Estimation of water quality model parameterswith simulated annealing algorithm

本文采用BOD—DO耦合模型模拟水质。在获取参数最优化值时采用BOD和DO耦合模型计算值与实测值之差的加权平方和作为目标函数,将模拟退火算法(SimulatedAnnealing,SA)应用到多河段水质模型的多参数的同时估值中。采用30次重复估值试验,并记录所有试验中最优估计值的最大值和最小值,所得到的各河段BOD衰减系数的最优估值与真值的误差和河水复氧系数的最优估值与真值的误差均在允许范围内。说明了方法的可靠性。

The simulated annealing (SA) algorithm is applied to simultaneously estimate the parameters of water quality in multiple river sections, and the BOD-DO coupling model is used to simulate the water qualities. In order to optimize the estimation of river water quality the difference of the sum of weighted square between predicted and measured BOD and DO is adopted as the target function. The maximum and minimum optimal estimation of the parameters are deduced after 30 times repeated estimation calculation. The result shows that the errors of estimated BOD and DO value are in the acceptable range. Accordingly, the optimal planning of water pollution control can be carried out.