具有最优模型传递矩阵的交互式多模型算法

摘要

在传统的交互式多模型（ IMM，interacting multiple model）算法中，描述模型马尔科夫切换过程的模型传递矩阵被定义成一个常值矩阵，并且将子滤波器间的相关性信息遗漏。然而，由于实际环境的复杂性，传统的IMM算法无法满足飞行器跟踪的需求。为此，提出一种具有最优模型传递矩阵的交互式多模型（ OMTM－IMM，optimal mode transition matrix IMM）算法，该算法在考虑子滤波器相关性的前提下，以线性最小方差理论为基础，推导出最优的模型传递矩阵，该传递矩阵更加符合实际情况，理论分析和仿真实验表明该算法有效地提高了飞行器跟踪精度。%The traditional interacting multiple model ( IMM) algorithm usually models the mode evolutions as Markov processes with constant mode transition matrix and leaves the correlative information among sub⁃filters out. However, because of the complexity of the practical application, the traditional IMM algorithm is unsuitable in aircraft tracking. To solve these problems, an optimal mode transition matrix IMM algorithm ( OMTM⁃IMM) is presented. The new algorithm uses the linear minimum variance theory to calculate the optimal mode transition matrix according to the correlations between sub⁃filters. In this case, the new matrix further approaches the truth one, and the estimation accuracy can be improved. This conclusion can be support by the following theoretical derivation and simulations in aircraft tracking.