基于Bayesian-Fisher混合模型改进的交互式多模型算法

An Improved Interacting Multiple Model Algorithm Based on Bayesian-Fisher Hybrid Model

针对交互式多模型算法(interacting multiple model,IMM)存在模型集设计困难的问题,提出一种改进的IMM算法。该算法将当前统计模型(current statistical,CS)融入到Bayesian-Fisher混合模型中,实现对加速度均值的在线自适应调整,从而提高对过程噪声协方差的估计精度,减少模型失配。同时,对匀速模型(constant velocity,CV)进行改进,并将改进的CS、CV模型在IMM算法的体系下进行交互。仿真结果表明,改进的IMM算法能快速响应目标状态的变化,取得优于IMMCVCA、IMMCVCS、IMMCVCACT以及IMMCVSTMIE的跟踪性能。

An improved interacting multiple model algorithm was proposed to solve the problem that the interacting multiple model (IMM) had difficulties in designing mode set.The algorithm integrated the current statistical model into the Bayesian-Fisher hybrid model to achieve the online adaptive adjustment of the mean acceleration,thereby improving the estimation accuracy of the process noise covariance and reducing the model mismatch.At the same time,constant velocity (CV) model was improved and the improved CS and CV model were interacted under the IMM algorithm system The simulation results show that the improved IMM algorithm can quickly respond to the change of target state and achieve better tracking performance than IMMCVCA,IMMCVCS,IMMCVCACT,and IMMCVSTMIE.