Research On Algorithms For Single Observer Passive Location And Tracking Based On Spatial-frequency Domain Information

Single observer passive location tracking system with simple equipment has become thefocus of research in the field of passive location, as it is relatively independent and highlymanoeuvrability. The single passive location method of emitter based on the observedinformation of spatial-frequency domain has broad prospects in application, because of its fastertracking velocity and higher location precision compared with the other traditional methods. Inorder to realize the goal of fast and high location and tracking in the field of single observerpassive location, some critical questions on land-based fixed single station are touched based onspatial-frequency domain information in this dissertation. This questions include the locationerror analysis, observability analysis and tracking filtering algorithms of constant velocity targetand maneuvering target. The main contents are summarized as follows:1. The location error analysis of the location method based on spatial-frequency domaininformation is touched. The various factors affect the location accuracy of the location methodand how these factors influence the location accuracy can be gotten by the location error analysis.Theoretical analysis and simulation results show that the method is most sensitive to angularvelocity measuring error. The tracking accuracy of the location method when tracking the near,fast moving and high radiation frequency target is high. But there will be non-observable regionswhen using the location method to track target.2. It studies the observability of the constant velocity target and maneuvering target usingthe spatial-frequency domain information. Observability analysis is the basis of location trackingalgorithm. The study on tracking algorithm is useful only when the target meet the observableconditions. As the observability of nonlinear system need calculation the complex Jacobianmatrix, a new analysis method is presented. It uses the pseudo-linear processing for themeasurement equation, and then uses the observability analysis of the theory of linear systems toanalyze the observable conditions of constant velocity target and maneuvering target. Themaneuvering target include constant accelerated motion and constant turn rate motion. As longas the target do not make radial movement toward the observer or circular motion around theobserver can be observed. The experimental results show that the conclusions are correct.3. As the calculation of unscented kalman filter is large, an improved filter algorithm basedon schmidt orthogonal transform is presented. The algorithm used schmidt orthogonal transformsampling strategy to reduce the number of sampling points. The computer simulation resultsshow that the algorithm can improve the calculation efficiency in the guarantee of filteringprecision and can be easily implemented in real time. Then the unscented kalman filter in singleobserver passive location is sensitive to the initial value and will divergent because of thenumerical calculation error, an improved forward-backward smoothing algorithm based onsquare-root unscented kalman filter is presented. The algorithm used the covariance square rootmatrix instead of covariance matrix in the process of estimation and utilized backwardsmoothing to get a more accurate state estimate as an initial condition. Simulation results showthat the algorithm improves the stability of the filter and the robustness of the initial value. 4. Through the simulation experiments of tracking different maneuvering target usedcommonly models such as constant velocity model, constant acceleration model, Singer modeland the current statistical model can know that the performance of the current statistical model isbest. As the tracking accuracy of the current statistical model when tracking maneuveringweakly target is low, an improved CS model is proposed. The improved model uses therelationship between the acceleration maximum value and the motion state of the target, so thatthe motion state of the target can reflect in the acceleration variance. The experimental resultsshow that the improved model have better tracking performance for maneuvering target.Meanwhile, in order to improve the tracking performance of the interacting multiple modelalgorithm on the target state mutation, a novel interacting multiple model algorithm is presented.The time-varying fading factor of the strong tracking filter is used to correct the prediction errorcovariance, so the gain can be adjusted in real time. The simulation results show that theproposed algorithm can improve the tracking performance for maneuvering target.