摘要
无迹卡尔曼滤波(UKF)算法在原先状态分布中按某规则取点,并使点的均值和协方差等于原状态分布的均值和协方差。将这些点代入非线性函数中,得到相应的非线性函数值点集,并通过点集求取变换后的均值和协方差。其应用步骤包括计算及传递取样点、利用预测取样点和权值计算预测均值和协方差、预测测量值和协方差,最后计算UKF增益,更新状态向量和方差。仿真表明该方法比EKF方法可用性更强。
According to some regulations, the unscented Kalman filter algorithm acquired points in the origin state distribution; the mean and the covariance of the points were equal to the mean and the covariance of the origin state distribution. Those points were used in nonlinear function to acquire the corresponding nonlinear function value point set; and the transformed mean and the covariance were acquired through point set. Its application process includes calculating and transmitting the mean and the covariance; making use of the forecast sample points and weighing calculation to forecast the mean and the covariance; forecasting measurement value and covariance; at last, calculating the UKF plus, renewing state vector and variance. The simulation showed that this method is more practical than EKF method.
出处
《兵工自动化》
2006年第8期73-75,共3页
Ordnance Industry Automation
关键词
卡尔曼滤波
无迹变换
无迹卡尔曼滤波
Kalman filter
Unscented transformation
Unscented kalman filter