A Kullback-Leibler-Based IMM Information Filter for the Jump Markov System with Unknown Noise

作者：Shen, Chen^*; Huang, Wei

来源：Circuits, Systems, and Signal Processing, 2018, 37(9): 4065-4081.

DOI：10.1007/s00034-017-0735-y

• Cubature information filter
• Interacting Multiple Model
• Kullback-Leibler divergence
• Jump Markov system

This paper is concerned with the state estimation of the jump Markov system with the unknown measurement noise. The proposed algorithm is derived under the framework of Interacting Multiple Model approach, and the recently reported Kullback-Leibler (KL) divergence-based scheme is used for estimation fusion. To facilitate KL divergence-based scheme, the information state and Wishart distribution are, respectively, used to describe the state and the unknown precision matrix of the measurement noise. Specifically, at both the mixing and estimation fusion stages, the KL divergence-based fusion scheme is adopted to fuse the information matrices, information state vectors, and parameters of the Wishart distribution from all the modes. At mode-conditioned filtering stage, parallel noise adaptive cubature information filters are designed to recursively estimate the information states, information matrices, and the Wishart distributed noise parameters. Simulation results prove the efficacy of the proposed approach.

• 出版日期2018-9
• 单位浙江工商大学; 中国舰船研究设计中心