摘要
根据UKF(unscented Kalman fliter)的工作机理,对所建立的双星(GPS、劳兰C)/航位推算(DR)舰船组合导航连续非线性系统模型进行解算,形成一套UKF在舰船组合导航中的递推算法。该方法根据随机变量的先验统计特性,按照特定的规则将状态变量分解成2n+1维的散布形式,然后利用统计线性回归技术,实现对非线性函数的线性化,可以获得更小的线性化误差。实船数据试验表明:UKF算法与扩展卡尔曼滤波(EKF)算法相比,系统稳定性更好,滤波器估计精度更高。
According to the mechanism of the unscented Kalman fliter（UKF）, a continuous nonlinear model of BeiDou（GPS/RLC）/DR integrated navigation system was established, and a corresponding algorithm was proposed. Based on prior statistic characteristic of random variables, state variables are disintegrated into 2n ＋ 1 samples in the light of specific rule, and the nonlinear function is linearized by statistic linear regressive technology. By this means, linearizati＇on errors are reduced effectively. The experiment with actual data from warship shows that UKF algorithm＇s estimation precision and the system stability are much better than those of EKF algorithm.
出处
《兵工学报》
EI
CAS
CSCD
北大核心
2007年第5期539-542,共4页
Acta Armamentarii
