Research On Underwater Target Tracking Method Based On Interactive Multiple Models In Complex Noise Environment

In order to make better use and development of marine resources,underwater target tracking technology has become a current research hotspot.Target tracking is to estimate the state of a moving object using the measurements available from various sensors and through filtering algorithms.In order to ensure better tracking of maneuvering targets underwater,more reliable and stable filtering algorithms are required.However,due to the special underwater environment,the system may have measurement outliers and inaccurate state noise,and the accuracy of the tracking system will be greatly affected at this time.Based on this background,this paper takes the underwater maneuvering target as the research object,and focuses on the two problems of measurement outlier interference and unknown state noise statistical information in the system.A stable and reliable filtering algorithm is designed and used for underwater maneuvering.target tracking system.First,the theoretical knowledge used in this article in underwater target tracking is briefly introduced,covering several classic target motion models and filtering algorithm,and introduced the Variational Bayesian(VB)method for dealing with non-Gaussian and imprecise noise problems.Finally,the basic principle of the IMM algorithm is introduced and a general analysis is carried out.Then,starting with the classic IMM algorithm,the IMM-based Kalman filter(IMMKF)algorithm is designed for the Gaussian noise environment to track underwater maneuvering targets.The impact of measurement outliers on target tracking performance is explained in detail.For the problem of measurement outliers,the classic Huber-based robust filter algorithm is introduced,and the defects of Huber method in IMM algorithm are analyzed through simulation.Then,in view of the outlier interference in the measurement information in the underwater tracking system,the method based on statistical modeling was used to model the measurement noise into the student t distribution,and the likelihood was re-derived in the VB framework function,designing a kind of student t robust filter based on IMM(IMM-RSTF),and the filtering algorithm is compared and analyzed by simulation.Finally,in view of the inaccuracy of the state noise model in the underwater tracking system,for the situation that the variance matrix of the state noise of the system is unknown and changes with time,IMM-based Sage-Husa adaptive student t robust filter(IMM-SH-RSTF)and IMMbased variational Bayesian adaptive student robust filter(IMM-VB-RSTAF),and the above filtering algorithm is analyzed through simulation,which proves that our proposed IMM-VBRSTAF algorithm has higher accuracy and provides a stable and reliable filtering method for the mobile target tracking system in the underwater complex environment.