Low-complexity ATPM-VSIMM algorithm with adaptive model parameters

Abstract

Aiming at the problem that for maneuvering target tracking, the accuracy of tracking degraded in interacting multiple model algorithms due to the fixed model sets and the fixed transition probability matrix, a low-complexity ATPM-VSIMM algorithm was proposed, which could update the model parameters adaptively.The maneuvering situation of the target was judged according to the innovation changes of the system, and the state noise of the model sets was adjusted to realize the adaptive update of the model sets.Then, the more accurate transition probability matrix was computed through the change of the model posterior probability and the inter-model switching relationship.Therefore, the matching degree between the system motion model and the target motion trajectory was improved.Finally, the high filtering accuracy and the fast response speed of the tracking system were guaranteed.The effectiveness of the proposed algorithm was verified through three aspects that are the initial value of the model posterior probability, the initial value of the transition probability matrix, and the state noise.Simulation results demonstrate that the filtering accuracy of the ATPM-VSIMM algorithm is improved about 8% compared with the existing algorithms.

Keywords

• maneuvering target tracking;adaptive state noise covariance matrix;adaptive transition probability matrix;variable structure interacting multiple model