Intelligent Marginal Particle Filter for Maneuvering Target Tracking

The particle filter (PF) has proven effective in maneuvering target tracking due to its capability to manage systems that are both nonlinear and non-Gaussian. However, challenges such as the curse of dimensionality and particle impoverishment are primary obstacles to applying PF in maneuvering target tracking. To tackle these issues, an intelligent marginal particle filter (IMPF) for maneuvering target tracking is proposed in this paper. The IMPF is formed by combining the intelligent particle filter (IPF) with the marginal particle filter (MPF). To reduce the dimension of particle filtering, the state vector has been initially decomposed into linear and nonlinear parts by the proposed filter. As a result, the linear part can be managed by a Kalman filter, and the nonlinear part is handled through particle filtering. Subsequently, genetic operators are introduced into the resampling process of particle filtering to adjust the small-weight particles. This action helps to alleviate particle impoverishment while introducing a moderate rise in computational demands. The outcomes derived from tracking maneuvering target experiments showcase the effectiveness of IMPF, especially when dealing with abrupt changes in the target’s state.