基于可信度优化的仿真模型参数估计方法  

Estimation method of simulation model parameters based on credibility optimization

作　　者：杨明[1] 焦松[1] 李伟[1] 陆凌云[1] 

机构地区：[1]哈尔滨工业大学航天学院,黑龙江哈尔滨150080

出　　处：《华中科技大学学报（自然科学版）》2014年第6期90-94,共5页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61273226);国家自然科学基金创新研究群体科学基金资助项目(61021002)

摘　　要：针对已知仿真模型参数取值范围而难以确定其精确取值的问题,在仿真模型验证工作的基础上提出了基于可信度优化的仿真模型参数估计方法.首先,在仿真模型上开展拉丁超立方试验,并评估得到每个试验点下的仿真模型可信度;然后,将所有试验点下的仿真模型参数和可信度作为训练样本,采用支持向量机得到仿真模型参数与可信度之间的关系模型,由此将动态的仿真模型可信度优化问题转变为静态的函数优化问题;最后,基于遗传算法得到使得仿真模型可信度最优的参数取值,即为仿真模型参数的估计值.在应用实例中,该方法有效地确定了某导弹导引系统仿真模型参数取值.The ranges of simulation model parameters can be obtained,but their accurate values are difficult to determine sometimes.To solve the problem,an optimization method of simulation model parameters based on the credibility optimization and the work of simulation model validation was proposed.The Latin hypercube experiment was carried out on the simulation model,and the credibility of simulation model was evaluated under each experimental point.Then,the values of simulation model parameters and credibility of simulation models under all experimental points were taken as training samples.The relational model between the simulation model parameters and the credibility were obtained using support vector machine.So,the dynamic optimization problem of simulation model parameters was transformed into a static optimization problem of function.Finally,the estimation values of simulation model parameters optimizing the credibility were obtained via the genetic algorithm. In the application,the accurate parameters values of the simulation model of a missile guidance system were determined effectively.

关 键 词：参数估计 可信度 仿真模型验证 支持向量机 遗传算法 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]