The covariance intersection fusion estimation algorithm weighted by diagonal matrix based on genetic simulated annealing algorithm and machine learning

This paper is concerned with covariance intersection (CI) fusion for multi-sensor linear time-varying systems with unknown cross-covariance. Firstly, a CI fusion weighted by diagonal matrix (DCI) is proposed, and it is proved to be unbiased and robust and has higher accuracy than classical CI fusion. Secondly, the genetic simulated annealing (GSA) algorithm is used for multi-parameter optimization problem caused by diagonal matrix weights. Considering the serious time-consuming problem in optimization process of the GSA algorithm, Back Propagation (BP) network is used to obtain the optimal weights. Eventually, the DCI based on GSA algorithm and BP network is proposed. The proposed algorithm has higher accuracy and better stability than classic CI fusion algorithms. Simulation analyses verify the effectiveness and correctness of the conclusion.

Automated summary

This paper focuses on covariance intersection (CI) fusion for multi-sensor linear time-varying systems with unknown cross-covariance. A CI fusion weighted by diagonal matrix (DCI) is proposed, which is proven to be unbiased, robust and more accurate than classical CI fusion. The genetic simulated annealing (GSA) algorithm is used for multi-parameter optimization problem caused by diagonal matrix weights. To address the time-consuming optimization process in the GSA algorithm, the Back Propagation (BP) network is employed to obtain the optimal weights. The proposed DCI based on GSA algorithm and BP network achieves higher accuracy and better stability compared to classic CI fusion algorithms. Simulation analyses validate the effectiveness and correctness of the conclusion.