一种改进的多模型噪声辨识方法

A Reformative Multiple Model Approach to Noise Identification

在混合系统参数估计问题中,交互式多模型算法是一种比较有效的求解方法。但是当系统机动模式较多时,模型集中的模型数目也随之增加。针对模型集过大导致的参数估计精度下降的现象,提出了一种改进的多模型算法,并将其应用于系统噪声辨识。它利用多扫描量测信息,采用离散优化技术,获得近似于系统的实际噪声水平的最优的子模型集。然后,利用此进行噪声估计作为结果输出。Monte Carlo仿真结果表明了新算法估计精度优于标准IMM算法。

The interactive multiple model algorithm (IMM) is an effective solution to the hybrid estimation problem. The model set augments along with system behavior patterns increase. A reformative multiple model approach is presented to improve the accuracy of parameter estimation when the model set is too large, and it is used to identify system noise. By means of the multiple scans of measurements information and the discrete optimization technique, the optimal sub-model set can be obtained which is close to the real noise level of the system. The final output is obtained by using it. The simulation results indicate that the new algorithm is better than the standard IMM.