一种INS性能参数重复性离群数据挖掘方法

Mining Method of Outlier Data for INS Performance Parameters Repeatability

关键词:  惯性导航系统  性能参数  重复性  离群数据

Key words:  Inertial navigation system  Performance parameters  Repeatability  Outlier data

摘要:

惯性导航系统参数长期重复性是影响导航性能的重要因素。批量生产的惯性导航系统在生产过程、运输存贮、标定等方面偶尔出现异常因素，导致个别产品性能参数长期重复性出现异常变化，进而影响惯导系统导航精度。为了快速挖掘异常数据，根据批次参数随时间变化特点，提出了一种多属性关联规则的惯性导航系统离群数据挖掘方法。通过对某型平台惯导系统参数长期重复性数据进行离群数据挖掘，结果表明对于参数长期重复性差导致的惯性导航系统性能异常现象, 使用所述方法可以有效检测出离群数据，并且能发现离群数据内部的关联关系。

Abstract:

The long-term repeatability of calibration parameters is an important factor affecting the navigation performance in inertial navigation system (INS). For the mass-produced INS, the abnormal factors in the production, transport storage, manual calibration, etc., result in the abnormal changes of the parameters long-term repeatability of individual production’s performance, which will affect the navigation accuracy of INS. In order to detect the abnormal INS performance according to the batched parameter changes over time , a multi-attribute association rule based outlier data mining method was proposed. And the effectiveness of this method was verified by outlier data mining about the long-term parameters repeatability of a certain type platform INS system.