Research On Active And Passive Location And Track Based On ESM And Feature Fusion

Target tracking are widely applied in military and civilian fields.Withthe development of electronic warfare technology, the traditional trackingmethods based on active sensors meet a challenge, but tracking methodsbased on passive sensors like ESM are getting more attention. For solvingthe problem of active and passive locating and tracking based on ESMpassive sensors，the main achievements in track field at home andabroad were researched and analyzed, two types of target trackingalgorithms based on data association and RFS（Random Finite Set） werestudied. The main contributions of the dissertation are as follows:1) ESM sensor was modeled; single/double ESM simulation softwarewas created to prepare for measurement platform of the follow-up trackingalgorithm research.2) Passive tracking of single target single ESM sensor scenario wasstudied. The UKF was combined with Gaussian mixture PHD filter to getUK-GM-PHD filter applicable to nonlinear conditions. The targetmeasurements were gained using single ESM sensor simulation software,the target was tracked by UK-GM-PHD algorithm. The simulations resultsshow that UK-GM-PHD algorithm is able to passive track a target undersingle ESM sensor.3) For solving the problem of coalesce neighboring tracks whentracking multiple targets which are spaced closely under high dense clutter,multiple targets tracking algorithm is proposed based on CKF and featureaided data association. The measurement is augmented by feature information and feature aid technology is integrated into Joint ProbabilisticData Association (JPDA).Then nonlinear measurements are processed andthe states of targets are updated by Cubature Kalman Filter (CKF). Theexperimental results show that the proposed algorithm improves thetracking accuracy and reduces the rate of lost target in comparison withtraditional JPDA algorithm.4) For solving multi-target passive tracking problem based on ESMsensors, the sequential UK-GM-CPHD based on track label was applied totrack targets.Firstly, the states of new targets and existed targets werepredicted one step by UKF, and then track labels were added to eachGaussian term, using GM-CPHD prediction and updating step, the finaltrack results was obtained through the track management. Simulationresults show that the sequential L-UK-GM-CPHD algorithm using doubleESM sensor platforms performs better tracking performance than theL-UK-GM-CPHD using single ESM sensor.