A Better Adaptive IMM(Interactive Multiple Models) Algorithm Based on A Posteriori Information Modification
Huang He,Wang Xiaoxu,Zhao Chunhui,Liang Yan,Pan Quan
DOI: https://doi.org/10.3969/j.issn.1000-2758.2011.06.001
2011-01-01
Abstract:Aim.The traditional IMM algorithm is,in our opinion,still deficient in the tracking accuracy and stability of a target tracking system with uncertain a priori information.We now propose what we believe to be a better adaptive IMM algorithm that can identify on line the Markov transfer probability parameters in the process of the Kalman filtering.In the full paper,we explain our algorithm in some detail;in this Abstract we just add some pertinent remarks to listing the subsections of section 1: the traditional IMM algorithm(subsection 1.1),the principles of the adaptive IMM algorithm based on a posteriori information modification(subsection 1.2),the design of the IMM algorithm(subsection 1.3) and the simulation results and their analysis(subsection 1.4).In subsection 1.1,we point out that the deficiencies of the traditional IMM algorithm are that the tracking is not adaptive and that the tracking precision is low.In subsection 1.2,we give Fig.1 to show the schematic of the adaptive IMM algorithm.In subsection 1.3,we design our IMM algorithm by using the principles presented in subsection 1.2,which adaptively modifies the a priori Markov transfer probability parameters in accordance with the changes in the error compression rate of a non-matching model,forgets the information on the non-matching model and expands the information on the matching model,thus greatly increasing the convergence speed of the target tracking system.In subsection 1.4,we perform the computer simulations of our adaptive IMM algorithm;the simulation results,given in Fig.2 and their analysis show preliminarily that our IMM algorithm outperforms the tracking accuracy of the traditional IMM algorithm.