交互式多模型算法性能分析

广泛应用于目标跟踪和故障诊断的动态多模型估计（SMME）假设模式的切换服从马尔可夫过程，对马尔可夫切换概率（MTP）的非MonteCarlo分析有助于深入了解SMME的机理，发现参数寻优的法则，设计或发展新的自适应多模型估计器，然而由于SMME的复杂性，非Monte Carlo分析很难给出，本文针对SMM中著名的交互式多模型（IMM）估计器，通过将IMM看作是输入交互和子滤波器串联，分析了具有m个参数的MTP矩阵，给出了六条不依赖主动应用环境及子滤波器设计的结论，部分结果也适用于一阶广义伪贝叶斯算法（GPBI）。

Analysis on the Performance and Properties of Interacting Multiple Models Algorithm

There is less nonsimulation analysis of markov transition probability (MTP) of switching multiple model estimation (SMME), which may be helpful to provide good insight into the essence of SMME, supply some rules for parameter design and develop more adaptive multiple model estimators. This paper presents a nonsimulation method to analysis MTP matrix with \%m\% prameters (\$m\$ is the model number) of interacting multiple model (IMM) estimator, which is well known in SMM. Several conclusions, independent of scenarios and model conditional filters, are given. Part of the analysis is also suitable for one order gereralized pseudo Bayesian (GPB1).