中文说明：应用背景以往的状态估计器设计一般是基于一个参数固定的系统模型，如若工作状态变化，往往会引起模型的不匹配而引入较大误差。而IMM算法使用2个或更多的模型来描述工作过程中可能的状态，最后通过有效的加权融合进行系统状态估计，这就很好地克服了单模型不准确的问题。关键技术在机动目标跟踪中，交互式多模型(InteractingMultiple Model, IMM)算法是一种很有效的方法。该方法的特点是通过马尔可夫转移概率对多模型进行切换，自动调节滤波带宽，能跟踪目标的任意机动。但其各模型滤波算法通常采用卡尔曼滤波或扩展卡尔曼滤波算法，对于非线性、非高斯系统模型，其滤波性能将大大降低。
English Description：
Application backgroundIn the past, the design of the state estimator is usually based on a system model of a parameter, and if the change of the working state often leads to the error of the model mismatch.. The IMM algorithm using two or more models to describe the working process. Finally, the weighted fusion system state estimation, which well overcomes the problem of single model is not accurate.Key TechnologyIn maneuvering target tracking, the interactive Model (IMM) algorithm is a very effective method.. The feature of the method is that the multi model is switched by Markov transfer probability, and the filter bandwidth can be adjusted automatically, and the target can be tracked.. However, the filtering algorithm of each model is usually used Calman filter or extended Calman filtering algorithm, for the nonlinear and non Gauss system model, the filtering performance will be greatly reduced.