关键词: 强非线性、非高斯过程噪声系统, 预测滤波, 高阶容积卡尔曼滤波, 预测-五阶容积卡尔曼滤波, 五阶球面单形积分准则, 广义-高斯拉盖尔积分准则

Abstract: In order to be suitable for strongly nonlinear and non-Gaussian process noise systems, combining predictive filtering (PF) and high-degree cubature Kalman filter (HCKF), a Method of predictive fifth-degree cubature Kalman filter (P5thCKF) is proposed. The prediction method was used to correct the system model in real time, and then the new model was substituted into the fifth-degree cubature Kalman filter(5thCKF) for real-time recursive state estimation. This paper derives the fifth-degree spherical simplex-radial integral criterion, uses the fifth-degree spherical simplex integral criterion to deal with the sphere area, and the generalized Gauss-Laguerre integral criterion deals with the radial integral; describes the predictive filtering method and derives the model error adjustment. Simulation results show that the proposed algorithm has higher filtering accuracy than volumetric Kalman filter and fifth-degree volumetric Kalman filter in nonlinear and non-Gaussian process noise systems.

Key words: Strongly nonlinear and non-Gaussian process noise systems, predictive filtering, high-degree cubature Kalman filter, predictive fifth-degree cubature Kalman filter, fifth-degree spherical simple integral criterion, generalized-Gauss Laguerre integral criterion