基于广义线性模型的变参数广义卡尔曼滤波

给出了连续系统的一种广义线性模型，不仅适用于已知系统，而且适用于未知系统，无论系统是线性还是非线性的。该模型系数服从确定的变化规律而无需在线辨识或预先计算，因而实时性好。定义了基于这种模型的变参数广义卡尔曼滤波。并给出其递推计算公式，可对未知系统实现实时滤波估计，且满足均方误差最小。该系统能方便、快捷、准确地实现未来多个时刻状态群的实时预报，以便提早掌握系统未来的运动趋势，提高系统的控制性能和对机动目标拦截的制导精度。算例仿真表明，结果是令人满意的。

Parameter Varying Generalized Kalman Filter Based on Generalized Linear Model

A generalized linear model is given,for systems containing linearity and nonlinearity,which may be used for both known and unknown continuous systems.The coefficients of the models are fixed without distinguished on-line or predefined,therefore,the real-time of the models is good.The generalized Karlman filter is also defined with this model,and the recurrence formulas are provided.The real-time estimation for the unknown system is completed with the minimum mean-square error.Because the groups of the states in the future is estimated conveniently,rapidly and accurately,so the move-tension of the systems in the future can be grasped early and both the function of the control systems and the guidance-systems-accuracy are improved.