基于期望模型的变结构交互多模型算法

Interactive Multiple-model Algorithm based on Expected Model with Variable Structure

采用变结构思想,提出一种基于期望系统噪声模型的变结构交互多模型算法。算法中,预测噪声变化范围的噪声模型集不再采用固定结构,而是根据量测新息在线辨识和调整,使之更加接近于目标的真实运动模式。基于模型匹配概率与先验的马尔科夫切换概率调整期望系统噪声模型噪声方差,得到一个最优的系统噪声模型集。将该算法与交互多模型算法相结合,得到一种变结构交互多模型算法。仿真结果表明,该算法对机动目标的跟踪精度要优于固定结构的交互多模型算法。

The adaptive maneuvering target tracking algorithm based on expected system noise model is improved based on the idea of variable structure.The model set that forecast the varying range of target＇s noise is no longer a fixed structure,and is identified and adjusted according to the measurement of innovation.Under this condition,the model set is more adjacent to the real motion model of target.The expected system noise model is adjusted based on model probability and Markov transferring probability.Together with the IMM,a adaptive IMM algorithm is formed.Through the Monte-Carlo simulation,this improved algorithm is superior to the unimproved algorithm and the standard IMM algorithm on performance of tracking maneuvering target.