Research Of Multisensor Maneuvering Target Tracking Information Fusion Algorithm

As an emerging overlapping course, sensor information fusion has a wide range of applications in many areas. And as application example of information fusion technology in target tracking area, multisensor target tracking combines multisensor information organically in order to improve estimation precision of target movement status. Compared with any single sensor target tracking, multisensor target tracking is more efficient and precise. At present, information fusion algorithm is one of the focuses in target tracking problems.The thesis focuses on two key questions in target tracking; these are status estimation and data association. On research of optimal and sub-optimal Kalman filter algorithms and estimating target status, it uses interacting multiple models, the typical algorithm which is used to deal with maneuvering target tracking, and multisensor joint probabilistic data association to track multi- maneuvering target. Then, maneuvering target tracking methods are simulated experimentally. This paper mainly concerns with the following contents.①Thesis discusses all kinds of optimal and sub-optimal filter algorithms, such as kalman filter, extended kalman filter and unscented kalman filter. Take the rocket for an example, because of measurement data containing outliers during actual flight of rocket and nonlinear measurement equation during tracking using radar and infrared sensors, the paper proposes unscented kalman filter algorithm with Outlier detection. The simulation results indicate that the new algorithm is efficient to rocket status estimation and it has a higher filtering estimation precision compared with extended kalman filter.②In order to solve the multi-target target problem, this thesis reveals parallel and sequential processing structure of multisensor joint probabilistic data association based on research of formation of tracking gate and sensor joint probabilistic data association algorithm. The simulation results verify its effectiveness under clutter.③Aiming at the uncertainty of maneuvering target movement model and existed clutter, based on the interacting multiple model, the paper brings in multisensor joint probabilistic data association method to accomplish association of multisensor measurement data with maneuvering target, and considering maybe nonlinearity of target model, our approach is presented. That is interacting multiple model multisensor joint probabilistic data association extended or unscented kalman filter algorithm. The simulation results indicate that the methods have better tracking results in the case of nonlinear movement model period. And it also tests that unscented kalman filter has smaller filtering error and higher estimation accuracy compared with extended kalman filter.At last, the thesis summarizes all research work discussed above, and points out the futher research direction in this field.