Radar Target Tracking Algorithm Research Under The Nonlinear Measurement

Target tracking has been widely applied in military and civilian field,in which filtering algorithm plays a core part.Since most of the real-time radar systems are set up in polar coordinates or ball coordinates,the Standard Kalman Filtering algorithm is no longer usable when dealing with the nonlinear system measurement equations.Then nonlinear filtering algorithms are invented and become popular in target tracking.To deal with this particular problem,this paper studied several filtering algorithms based on non-linear measurement information and its application in radar target tracking.First of all,this paper reviewed several kinds of classic nonlinear filtering algorithm,including Extended Kalman Filtering(EKF),Unscented Kalman Filtering(UKF),Particle Filtering(PF)algorithm.The performance of these three algorithms were evaluated and compared from the root mean square error and the computational complexity through simulation.The second,under the precondition that one can only use the radar measurement system to obtain the target position measurement information,we studied the existing Converted Measurement Kalman Filtering(CMKF)algorithms,including Unbiased CMKF,Modified Unbiased CMKF and Decorrelated Unbiased Converted Measurement Kalman Filter Based on the predicted position(DUCMKF).On the base of it,simulation was made to compare all those three algorithms.Then we went to the scenario which happens a lot in real-time tracking cases that provided us more usable information about target’s doppler velocity.First we introduced two typical algorithms to solve the radar target tracking problem with range-rate measurements.One is sequential converted doppler measurement kalman filter,another is statically fused converted position and doppler measurement kalman filters.Then we pushed DUCMKF further and proposed sequential decorrelated unbiased converted position and doppler measurement kalman filter based on the predicted position(SQ-DUCMKF).Still,simulation was made to prove the new algorithm,weighing multiple factors in reality,can archive more accuracy and only add little computation.Finally we studied the best linear unbiased estimator.First we introduced BLUE,then through a series of equivalent transformation steps we deduced the BLUE algorithm in Kalman frame structure.It reminded me that we could develop a similar algorithm like BLUE using SQ-DUCMKF’s architecture,extending BLUE to the use of Doppler information.Simulation showed that BLUE and DUCMKF were equal in performanc,so were SQ-DUCMKF and the sequential best linear unbiased estimator(SQ-BLUE).Both SQ-DUCMKF and SQ-BLUE can achieve high-precision target tracking in position measurement information as well as with Doppler measurement information.