一类带乘性噪声多通道非线性系统的平滑算法

Smoothing Algorithm for a Class of Multi-Channel Nonlinear Systems with Multiplicative Noise

在各通道乘性噪声不同的情况下,针对多通道带乘性噪声非线性系统的状态估计问题,提出1种状态平滑算法。该算法运用扩展卡尔曼滤波方法先根据全部观测数据对状态进行滤波估计,并存储一步预测估计值和一步预测估计误差的方差,利用存储的数据进行递推运算,得到状态的固定域平滑估计。仿真结果表明平滑算法较滤波算法精确性更高,稳定性更强。

This paper proposes a state smoothing algorithm to solve the problem of state estimation of multichannel nonlinear systems with multiplicative noise when the multiplicative noise is different in each channel. Using the extended Kalman filter, this algorithm first conducted filtering based on all the observation data, saving the results of both one-step prediction and square of one-step prediction error, and then conducted recursion with the saved data. Finally, the state estimation in terms of fixed field smoothing was obtained. The simulation results show that the smoothing algorithm is more accurate and more stable than the filter algorithm.