Hummingbirds optimization algorithm-based particle filter for maneuvering target tracking

Abstract

As a commonly used filtering method for nonlinear non-Gaussian systems, particle filters (PFs) have been successfully applied in the field of maneuvering target tracking. However, particle impoverishment is a major obstacle to the PF performance. To overcome this defect, this paper combines the hummingbirds optimization algorithm (HOA) with a standard PF and proposes an HOA-based PF (HOA-PF) for maneuvering target tracking. The proposed filter treats the particles as individual hummingbirds, simulates the honey-collecting process of hummingbirds in nature and moves the particles as a whole to the high-likelihood region by performing self-searching and guided-searching phases. Moreover, to enhance the particle diversity, the mutation method of the following birds in the HOA is improved. Thus, the distribution of particles in the HOA-PF is reasonable. The results of experiments on the univariate nonstationary growth model and the maneuvering target tracking problem demonstrate the effectiveness of the proposed method.