Research On Key Technology Of Maneuveirng Target Tracking

With the rapid development of the technology and science, the environment of themodern war has been more and more complex. The maneuverability of targets is becomingmore and more complicated and inconstant; meanwhile the demand of the trackingperformance index also is increasing. Hence, the targets tracking technology becomes theresearch hinge in the military and civil. In the targets tracking technology, the state optimalestimation algorithm is the mathematical tool and guarantee for realizing the trackingtechnology. The modeling of the target is the foundation and the kernel of the trackingsystems. The system configuration design is the effective mean to implement the trackingalgorithm. Therefore, this dissertation studies for the maneuvering target tracking from thethree statements above, which can be summarized as follows:The nonlinear character of the modern targets tracking system has been analyzed. Thenonlinear filter methods that include Extend Kalman Filter (EKF), Unscented Kalman Filter(UKF) and Cubature Kalman Filter (CKF) are elaborated in theory. Then the precision andthe stabilization performance of state estimation in the high dimensional situation wasanalyzed and verified by simulation, which provides the theoretical basis for the adaptivealgorithm research on the nonlinear maneuver targets tracking system subsequently.Then theoretic analyzing and performance of the current statistical model and itsadaptive maneuver targets tracking was elaborated and the problem that the maneuveringfrequency of the targets maneuvering was hardly defined accurately and the value can adjustwith the change of the targets maneuvering was point out, which caused the currentstatistical model can’t describe the target effectively and decrease the precision of targettracking systems. Then a maneuvering frequency adaptive algorithm based on maneuveringdetection is presented. The model of maneuvering frequency adaptive adjusted wasconstructed. Then the algorithm performance was analyzing and verified by simulation withtradition algorithm as the targets acceleration change repeatedly.The description of the acceleration of the target in the current statistical model wasmodified Raleigh distribution, which made the model has poor performance on weak andnon-motorized maneuvering targets. Then an adaptive filter algorithm based onextremum adaptive of acceleration was proposed, which a bell shape function as fuzzymembership function was utilized to adjust the upper and lower limits of target acceleration.At the same time, to enhance the response capability of the model as sudden maneuver or the acceleration changed greatly, a fading factor was proposed to adjust revised extremevalue of acceleration. The analysis and simulation results show that the algorithm has abetter performance on tracking weak and non-maneuvering maneuvering targets than thetraditional algorithm.The theory of target tracking and its performance based on Jerk model wasexpatiated and then that the system will have non-zero steady state error as the input signalof the target acceleration rate changed by jump function. That greatly reduces the trackingperformance of the Jerk model. An adaptive algorithm based on the MEP-on Jerk model formaneuvering target tracking then was proposed.By using the model prediction filter topredict the estimated model error of Jerk model, the influence caused by not fully estimatethe Jerk model is corrected online. At the same time the CKF was used to estimate the statefor nonlinear target tracking system. Analysis and simulation show that the algorithmperformance and the traditional algorithm as the target running with the non-motorizedsports, conventional turning movement and acceleration of the rate of change of the stepmaneuvers.Then the paper pointed out that the frequency and variance of maneuvering need topre-define and the value won’t be changed when the target maneuvering that makes theeffect of tracking drop down dramatically. A Jerk parameters adaptive estimation algorithmbased on the AR model was proposed. According to the parameter estimation method of theAR model with the MS model, the Jerk maneuvering frequency and variance of targetestimates real-time. At the same time as the CKF has complex account and sometime evenmaking the system divergence, the SRCKF was utilized. The analysis and simulation showthat the performance of this algorithm when the target running in the Jerk motion, comparedwith the Jerk model based on EKF algorithm and SRCKF algorithm.The general multiple model theory and the interacted multiple model theory wereintroduced. Then the Likely-Model Set algorithm (LMS) and the Expected-ModelAugmentation (EMA) of variable interacted multiple model was illuminate particularly. Asthe LMS and EMA all can’t cover all the real motion of target, a algorithm based onKullback-Liber (K-L) to analyze the model match degree with the others. Based on this, themodel set adaptive adjustment was implemented. As the model probability and thetransition matrix of model sets was changed in variable multiple model algorithm, a modelprobability updated algorithm was proposed and the equations to compute elements of thetransition matrix was deduced. The analysis and simulation show that the performance ofthis algorithm as the parameters changed all the time in target tracking.