Font Size: a A A

Research And Simulation Of The Key Models Of Radar Data Processing

Posted on:2011-06-06Degree:MasterType:Thesis
Country:ChinaCandidate:L ShenFull Text:PDF
GTID:2218330341951725Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
This thesis researched the simulation and validation of radar data processing, investigated the requirements of the establishment of simulation model base combined with the process and characteristics of radar data processing. It also classified the radar data processing module into three main categories (single target tracking, multi-target tracking and networking fusion tracking). It simulated and validated some radar data processing models and mainly analyzed the real time property and the validity of various algorithms. Finally, based on the recorded data of hardware-in-loop radar, the design and analysis of real time networking data processing fusion systems are proposed. Some significant conclusions are drawn. The main work can be summarized as follows:In the first section, the systematic summarization, simulation realization and performance analysis of single target tracking filtering algorithm are given. The motion models taken into consideration are uniform rectilinear model, uniformly accelerated rectilinear model, midcourse trajectory motion model and the Singer model. The algorithms included are the linear Kalman filtering algorithm, nonlinear filtering algorithm with the extended Kalman filtering (EKF) and interactive multi-model (IMM) filtering algorithm used in maneuver target tracking representatively. With aeroplane targets and trajectory targets as examples, the filtering performance of various filtering algorithms is compared. The conclusion is drawn. For aeroplane targets, current statistics model filtering and IMM filtering are preferred to ensure certain maneuver target tracking capacity; for trajectory targets, nonlinear filters based on kinematics such as EKF are preferred to ensure tracking accuracy.In the second section, the systematic summarization, simulation realization and performance analysis of track association and track management methods of multi-target tracking are given. It is shown in the simulation that the performance of elliptical window and the performance of rectangular window are equivalent and both are worth choosing. Then the performance of nearest-neighbour association algorithm and the performance of probability data association algorithm are compared. It is demonstrated in the simulation. For dense clutter, errors are likely to occur when nearest-neighbour association algorithm is used, but the performance of probability data association algorithm is stable. In addition, multi-dimensional probability data association (MPDA) algorithm is introduced, and the validity of association algorithm is checked by two different simulation situations (little space parallel track and uniform small angle cross track).In the third section, a further study on networking data processing model is given. With the distributed networking taken as an example, the simulation approach is employed to study the influence of such factors as radar station location, radar measurement error and radar networking strategy on networking fusion accuracy. The conclusion is drawn that the networking with the employment of three radars with similar measurement accuracy can improve fusion accuracy, and the employment of radar space diversity can also improve fusion accuracy. The accuracy of the networking of different types of radar is not improved obviously, and its fusion accuracy is approximate to the radar with the highest accuracy. The fusion accuracy of the networking with three radars is much better than that of the networking with two radars.Finally, based on hardware-in-loop radar equipment, the design and analysis of real time networking data fusion processing systems are made; communications operation and arithmetic operation are estimated; some significant conclusions are drawn. It is shown in the study that, for the distributed networking, temporal registration and track association are key factors that influence the real time property, and the influence of other factors such as coordinate transfer, communications delay and track fusion time is relatively small.The research results of this paper can provide a technical base for the establishment of simulation model, offer a framework of reference for the realization of real time data processing systems. It can be useful in military and engineering practice.
Keywords/Search Tags:Data processing, Multi-target tracking, Nonlinear filtering, Data association
PDF Full Text Request
Related items