GNSS/INS组合系统的抗差卡尔曼滤波

Robust Kalman Filtering of GNSS/INS Integrated System

为了有效探测和修复组合系统中存在的缓慢增长误差和异常值,从连续时间系统模型入手,推导出卡尔曼模型的动力学状态方程,并给出组合系统的状态方程和观测方程.在分析粗差对卡尔曼模型影响机理的基础上,得出抗差卡尔曼模型.针对传统抗差卡尔曼模型抗差参数需要在一定范围内主观确定的缺点,提出一种根据给定的错误预警率和故障探测率确定抗差参数的改进方法.利用改进方法对成熟软件产生的全球导航定位系统/惯性导航系统(GNSS/INS)的模拟数据和MonteCarlo方法产生的模拟单、多和缓慢增长误差模型进行实验研究.实验结果表明:用给定错误预警率和故障探测率确定抗差参数的抗差卡尔曼模型能够有效修复单或多异常值误差和缓慢增长误差,模型通过卡方检验.

To effectively detect and repair existing outliers and slowly growing errors（SGE） in the integrated system,the dynamic state equation and observation equation of the integrated system are given after the dynamic state equation is derived from continuous-time system model.By analyzing the mechanism of errors on kalman filtering model,a new robust kalman filtering model is proposed.The robust parameters of conventional kalman filtering is subjectively determined,so we introduce a new improved method according to the given alarm rate and fault detection rate.After the generation of GNSS/INS by the existing software and the various outliers by Monte Carlo method,the improved method is tested and the results show that the method effectively repairs the outliers and slowly growing errors by the given alarm rate and fault detection rate.Meantime,the model passes the χ2 test.