以模拟退火算法为核心着重讨论了水质模型参数的非数值随机优化方法。实例分析表明,利用非数值随机优化方法(包括模拟退火算法和遗传算法)对水质模型参数进行估计,可以获得较为理想的结果。不同参数估计方法的比较进一步阐述了非数值随机优化方法在参数估计中的优点

This paper focuses on stochastic parameter optimization for water quality model with simulated annealing algorithm (SA) which is discussed in detail. For comparison, genetic algorithm (GA) and steepest decent algorithm (SD) are also discussed. Simultaneously, the typical S P water quality model is adopted in a case study. Result of the case study shows that the stochastic optimization methods (SA and GA) are more effective than the other methods such as the steepest decent method. What are testified include not only in the aspect of theory but also in the case study, both SA and GA are able to reach the global optimal results. However, concerning SA and GA, GA is weaker in local optimization and spends more time in parameter optimization.