基于误差传递建模的仿真可信性控制  

Simulation credibility control based on error transfer model

作　　者：陈永兴[1] 吴晓燕[1] 滕江川[1] 吴静[2] 

机构地区：[1]空军工程大学防空反导学院 [2]中国人民解放军94221部队

出　　处：《计算机应用研究》2013年第7期1964-1966,1987,共4页Application Research of Computers

基　　金：陕西省自然科学基础研究计划资助项目(2012JM8020)

摘　　要：针对模型误差对仿真可信性的影响,提出了基于误差传递建模的仿真可信性控制方法。在建立仿真误差数学描述的基础上,利用泰勒展开方法对误差传递模型进行了推导,得到了仿真误差模型的一般形式。针对误差传递关系的非线性特点,提出了基于神经网络的误差传递模型参数辨识的方法,对神经网络相关参数的确定方法进行了研究,给出详细的辨识过程。为提高仿真可信性控制的效率,提出了基于敏感性分析的仿真模型综合重要性确定方法,最后,通过实例分析表明所建误差传递模型的正确性,以及该可信性控制方法的可用性和有效性。With respect to the relationship between model error and simulation credibility,this paper proposed a simulation credibility control method based on error model.Firstly,it gave the error description.It deducted the common form of error transfer model by Taylor expansion.Given to the nonlinear relationship of error transfer,it proposed a method based on neural network to identify the parameters of error transfer model.It gave the detailed process determine the parameter of neural network.Aiming at improving the efficiency of credibility control,it gave a method based on sensitivity analysis to calculate the importance of submodels.Finally,it gave an example prove the correctness of this error model and the usability of this credibility control method.

关 键 词：仿真 可信性控制 误差传递 神经网络 敏感分析 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]