基于UT变换的单步滞后无序量测算法

Single-step-lag OOSM algorithm based on unscented transformation

针对滞后无序量测(OOSM)的单步滞后滤波问题,在现有算法的基础上,推导非线性单步滞后无序量测更新方程.提出用UT变换来计算其中涉及到的状态向量以及相关量测之间的协方差,从而有效解决了状态转移方程为线性而量测方程为非线性的非线性Gauss系统的单步滞后OOSM问题.然后,针对多传感器单步滞后OOSM情况,给出了基于UT变换的单步滞后OOSM融合方法.与基于扩展Kalman滤波(EKF)框架下的EKFA1算法和不存在滞后情况的最优算法相比,新算法具有如下特点:可以适用于非线性量测方程的雅可比(Jacobian)矩阵或Hessian矩阵不存在的情况,具有较好的滤波性能,时间复杂度与EKFA1算法处于同一数量级.

Aiming at the out-of-sequence measurement （OOSM） problem, the update equations of the nonlinear single-step-lag OOSM are derived based on the existing methods. By introducing the unscented transformation （UT）, the covariance between state vector and corresponding measurement vector are computed such that the single-step-lag OOSM can be effectively solved under the nonlinear Gaussian system with nonlinear measurement equation and linear dynamic equation. Furthermore, a single-step-lag OOSM fusion algorithm based on UT is presented to confront the problem of the single-step-lag OOSM in multi-sensor system. The proposed algorithm has some advantages over the EKF A1 based on the extended Kalman filter frame and the optimal method without lags. For example, it can be used when the Jacobian matrix or the Hessian matrix of nonlinear measurement equation is nonexistent; its filtering performance is better; and its complexity has the same order of magnitude as that of the EKF A1 algorithm.