Adaptive Α-Β-γ Tracking Filter Restraining Outliers Based on Innovation Orthogonality

HUANG He,ZHANG Hui-sheng,HUANG Ying,HUI Xiao-bin,XU Jia-dong

DOI: https://doi.org/10.16182/j.cnki.joss.2010.04.020

2010-01-01

Abstract:Aim. The traditional α-β-γ filter is, in our opinion, still deficient in tracking targets with continuous outliers or of high maneuverability. A better adaptive α-β-γ filter that can apply innovation orthogonality to restraining outliers and fuzzy sets to making both tracking accuracy and convergence rate acceptable for such targets was proposed. In some detail, the better target tracking algorithm was explained. The judgment method of monitoring if the innovation orthogonal values of α-β-γ filter were lost or not and weight on the innovation sequences with activation function to inhibit the effect of the outliers was used. And five fuzzy subsets were used to represent input and output and the fuzzy subsets correspond to five functions of degrees of membership, as shown in Fig. 3 in the full paper. The output of the fuzzy logic system with the tracking error vector of better α-β-γ filter was adjusted according to the variety of locomotive target. Then the adaptive coefficients can be got so that the tracking results converge quickly on the premise of assuring the tracking accuracy. The analysis of computer simulation results, shown in Fig. 2 and Fig. 4, indicates preliminarily that the better tracking algorithm does restrain outliers and ensure that both tracking accuracy and convergence rate can be acceptable for targets of high maneuverability.