一类加权全局迭代参数卡尔曼滤波算法

A weighted global iteration Parametric Kalman Filter Algorithm

结合参数卡尔曼滤波算法和全局迭代推广卡尔曼滤波算法本文提出了加权全局迭代参数卡尔曼滤波算法。参数卡尔曼滤波算法可避免系统参数和状态变量之间的非线性耦合 ,同时通过带有目标函数的全局迭代算法保证能够获取到稳定、收敛的识别结果。分别针对线性结构模型和随动强化双线性结构模型进行了仿真参数识别。结果显示 ,不加权的全局迭代参数卡尔曼滤波算法对线性系统是有效的 ,而对非线性系统必须使用加权的全局迭代参数卡尔曼滤波算法。当信噪比较大 ,迭代无法得到收敛的结果时 。

By combining Parametric Kalman Filter Algorithm with weighted global iteration procedure, a weighted global iteration parametric Kalman Filter Algorithm(PKF\|WGI)is proposed . PKF algorithm can avoid the nonlinear coupling phenomenon between system parameters and state variables, and WGI procedure with an objective function is applied to obtain the stable and convergent solutions. The identification problems are investigated for single degree of freedom linear system and bilinear hysteretic systems. Ac cording to the numerical results, PKF\|WGI of weight 1 (i.e.: WGI without weight) is effective for the identification of linear system. While, An appropriate weight should be chosen to obtain good results for the identification of nonlinear system. When noise level is high, WGI with an objective function can ensure stable and convergent results.