基于非等维状态的IMM混合估计方法

在目标跟踪领域，交互多模型（IMM）估计器具有良好的性能和较低的复杂度。IMM的成功归因于模式混合，其中各模型输出用于模型条件重初始化。针对IMM算法中存在的非等维状态混合估计问题进行了研究，在总结现有算法的基础上提出了一种最优的IMM混合估计方法。该方法通过将"切换"态的概念引入目标状态，根据当前滤波时刻的模型概率和新息，动态地调整混合策略以实现最优估计。最后，通过仿真实验验证了所提算法在不同模型混合场景中的表现要优于现有的算法。 

IMM mixing estimation method based on unequal dimension states

The interacting multiple model (IMM) estimator has been proven to be of excellent performance and low complexity in tracking agile targets. The success of IMM attributes to mode mixing, where model outputs are mixed for model-conditional reinitialization. The problem of unequal dimension states mixing in IMM estimation is studied and an optimal method for IMM mixing is proposed based on summarizing the existing methods. By introducing the concept of "switching" state into the target state, the new method dynamically adjusts the hybrid strategy with model probability and innovation to achieve optimal estimation. The simulation results show that the proposed approach outperforms the existing algorithms in the scenarios of mixing different models.