摘要
为研究模型异参同效问题产生的原因,基于SCEM-UA算法,选用似然函数L(20)=(σ2ε)-20,在实际资料及理想资料两种情况下,分析了Nash模型参数率定中的异参同效问题。结果表明,在理想资料情况下,采用SCEM-UA算法能得到参数真值,即避免了参数的异参同效性影响,可见模型异参同效的根本原因是参数之间的互补作用或相依性,而精确的输入、输出和模型结构是参数率定的关键影响因素,优化方法只有在模型结构和资料均准确的条件下才能发挥其功效,从而避免了异参同效问题。
For study of parameter equifinality causes of hydrological model, SCEM-UA methodology with likelihood function L(20) : (σ t 2) -20 is used to analyze parameter identification of Nash model with actual data and ideal data. The results show that under condition of ideal data, the parameter of truth value will be obtained by SCEM-UA methodology which means that the dependence of the parameters is avoided and is the primary cause of parameter equifinality; the ac curate input-output and model structure is the key factors for influencing of parameter identification; optimization algo- rithm will only play role in parameter identification when the model structure and data are accurate, which the dependence of the parameters will be avoided.
出处
《水电能源科学》
北大核心
2014年第3期80-84,18,共6页
Water Resources and Power
基金
中南民族大学中央高校基本科研业务费专项资金项目(CZQ13006)
中南民族大学大学生创新创业训练计划(GCX13103)