The Research Of Algorithms For Aeromarine Single Observer Passive Location

With the fighting for maritime rights more and more intensely,maritime conflicts become increasingly frequent.The ability of detecting the target and estimating its accurate location to destroy the target becomes very important.Single observer passive location and tracking system has attracted extensive attention because of its advantages such as easily to fit,the large detection range and the strong capability of anti-reconnaissance.This paper studied single observer passive location technology in the environment of naval warfare.And,to the existing problems of the traditional positioning filtering algorithm such as low accuracy,slow speed and poor stability,the corresponding innovative methods were put forward,and the simulation results have proven their effectiveness.In chapter 2,the basic theory of single observer passive localization filtering algorithm was introduced,according to the latest results of the research,the location method using the direction of arrival,angular rate and Doppler frequency rate-of-change was analyzed.And the location error and influence factors of the method were also analyzed and summarized.In chapter 3,the selection,the gain-way and the accuracy range of the localization filtering algorithm's parameters were discussed,which give an applying foundation for the subsequent location filter algorithm.Considering the great uncertainty of the initial state and stronger nonlinear characteristics of single observer passive location system,the traditional Kalman filter and particle filter algorithms have an insufficient performance,such as poor stability,slow positioning speed and bad positional accuracy.To these problems,the research to improve the location accuracy,computing speed and robustness of the algorithms has been carried out,the specific work includes:According to the Unscented Kalman filter(UKF)algorithm,three kinds of improved algorithm have been proposed in chapter 4.Square root unscented Kalman filter location algorithm based on singular value decomposition(SVD-SRUKF)was proposed firstly,to solve the problem that the error of covariance matrix may become negative definite and even lead the filter divergent,SVD-SRUKF algorithm using singular value decomposition instead of Cholesky decomposition,which can insure that the filter with the covariance negative still keep high stability;A simplified iterative UKF algorithm was proposed secondly.This algorithm can make full use of the observation information through finite iteration to improve the convergence speed and accuracy,at the same time,because of the simplifying of sample points,this algorithm can reduce the running time and improve its efficiency;A robust iterative H-infinity filtering algorithm,using the strong tracking theory and combining H-infinity filter with UT transform,was proposed thirdly.The algorithm,using the advantage of H-infinity filter in noise control and the iterative theory,can reduce the noise influence on the output as far as possible,so it has a strong robustness.In chapter 5,to solve the degradation problem of the particle filter in the single observer passive location,considering the selection way of the important density function,an adaptive fading center differential particle filter(AFCDPF)algorithm was proposed.The algorithm,can optimize the system model by the adaptive fading factor,and use the filtering results as the important density function of the particle filter,then it provides more accurate initial sample for particle filter,so it effectively improves the accuracy of the positioning algorithm.To the problem that the traditional particle filtering algorithms need a large number of computing,using a control mechanism of particle number in AFCDPF algorithm can reduce the number of the particles in the filter and improve the operation efficiency?The particle filter algorithm was extended from the assumption of Gaussian noise background to the application of non-Gaussian noise environment,and two improved particle filter algorithms were proposed:an improved MCMC particle filter algorithm of positioning based on optimizing the initial sampling(IMCMCPF)and an improved Quasi Monte Carlo particle filter algorithm for positioning(IQMCPF).Firstly,in order to reduce the noise impact,the initial particle distribution of the two algorithms was optimized.And then,the efficiency of the two algorithms were improved by reducing the redundant particles for IMCMCPF and optimizing the selecting way of particles for IQMCPF,the practicability of the two algorithms were enhanced.