多元不确定仿真结果验证方法研究

发布时间：2018-06-18 04:57

本文选题：仿真结果验证 + 不确定性　；参考：《哈尔滨工业大学》2016年硕士论文

【摘要】：不确定性在仿真系统中广泛存在,在仿真领域引起越来越多的关注,研究输入参数包含不确定的仿真系统的结果验证具有实际需求。如何针对包含异类不确定性、输出为动(静)态形式的多元系统进行仿真结果验证,具有重要意义。本文围绕该问题展开下述研究。首先,针对静态输出,提出基于改进的Jousselme证据距离的仿真结果验证方法。通过分析静态输出特点,提出将其转化到证据理论中,采用改进的Jousselme证据距离度量证据体间的差异性,进而得到仿真系统的验证结果。其中,通过研究分析Jousselme证据距离及其修正证据距离存在的问题,提出了改进的Jousselme证据距离。其次,针对动态输出,提出基于改进的时间序列误差评估的仿真结果验证方法。针对动态输出特点,采用改进的时间序列误差评估方法将动态输出转化为相位、幅值和拓扑静态误差,进而转化为上述静态输出形式的仿真结果验证问题,运用改进的Jousselme证据距离度量得到仿真动态输出与实际动态输出间的一致性。通过改进相位误差评估算法,改进时间序列误差评估方法,扩展其应用范围。然后,针对多元输出仿真系统,提出采用主成分分析和区间主成分分析进行仿真结果综合验证。针对同时存在偶然不确定和认知不确定的仿真系统,采用区间主成分分析法,以降低输出维数和消除输出间的相关性;针对只存在偶然不确定的仿真系统,则采用主成分分析法。最后,结合本文研究的理论成果,设计开发了仿真结果验证辅助工具。从主界面、辅助工具功能模块和数据库设计三方面阐述了辅助工具开发过程,并将其应用于某飞行器末制导系统的结果验证。
[Abstract]:The uncertainty exists widely in the simulation system and attracts more and more attention in the field of simulation. It is necessary to study the results of the simulation system with uncertain input parameters. It is of great significance to verify the simulation results for multivariate systems with heterogeneous uncertainties and dynamic (static) output. This paper focuses on the following research. Firstly, a simulation method based on improved Jousselme evidence distance is proposed for static output. By analyzing the characteristics of static output, it is put forward that the improved Jousselme evidence distance is used to measure the differences of evidence bodies, and then the simulation results are obtained. By studying and analyzing the existing problems of Jousselme evidence distance and its modified evidence distance, an improved Jousselme evidence distance is proposed. Secondly, a simulation method based on improved time series error evaluation is proposed for dynamic output. According to the characteristics of dynamic output, the improved time series error evaluation method is used to transform the dynamic output into phase, amplitude and topological static error, and then to verify the simulation results in the form of static output mentioned above. Using the improved Jousselme evidence distance metric, the consistency between the simulation dynamic output and the actual dynamic output is obtained. By improving the phase error evaluation algorithm and improving the time series error evaluation method, its application range is expanded. Then, the principal component analysis (PCA) and interval principal component analysis (PCA) are used to verify the simulation results for the multivariate output simulation system. The interval principal component analysis method is used to reduce the output dimension and eliminate the correlation between outputs for the simulation system with both accidental uncertainty and cognitive uncertainty. Principal component analysis was used. Finally, combined with the theoretical results of this paper, an auxiliary tool for verification of simulation results is designed and developed. The development process of auxiliary tool is described from three aspects: main interface, function module of auxiliary tool and database design, and it is applied to verify the result of an aircraft terminal guidance system.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：V211;TP391.9

【相似文献】