Research On Fusion Algorithms Of Track Creating In Maneuvering Target Tracking

With the development of computer technology, communication technology, microelectronic technology and increasing complexity of modern wars, more and more multi-sensor systems are used in complicated applications, which forces people to integrate multiple sensors and different sources more effectively to enhance automation degree of data processing. As a recently growing up research front with broad prospects, information fusion is applied widely in multi-sensor information processing. Advantages of information fusion include: (1) different characteristics of a variety of types of sensors can be utilized comprehensively; (2) attributes of targets can be accessed from multi-aspect; (3) time and space coverage area of C3I system can be extended; (4) efficiency in the use of intelligence can be improved and credibility of information can be increased; (5) object detection probability and object recognization ability can be improved; (6) real-time situation assessment and threat assessment can be realized. Multi-sensor information fusion is in essence uncertain information processing. Fusion algorithms are mathematical methods to process uncertain information. As a hot spot of multi-sensor information fusion research, fusion algorithms have been attractive and many achievements have been obtained. With a multi-radar system under complex tracking environment as application background, measurement association algorithm, interacting multiple model algorithm and asynchronous track fusion algorithm of track creating in maneuvering target tracking have been studied.1. Measurement association algorithmFirstly, for single sensor measurement association problem, measurement assignment based data association (MSDA) algorithm and target reuse based data association (TRDA) algorithm are proposed. Secondly, on base of MSDA algorithm and TRDA algorithm, MMSDA algorithm, S-MSDA algorithm, MTRDA algorithm and S-TRDA algorithm are composed to process multiple-sensor measurement association. Finally, association performance of single sensor data association algorithms (including MSDA algorithm and TRDA algorithm) and multiple-sensor data association algotithms (including MMSDA algorithm and MTRDA algorithm) is compared in simulations.2. Interacting multiple model algorithmFor optimal construction of filtrering function set in interacting multiple model (IMM) algorithm, model set switching based multiple model (MSIMM) algorithm is proposed. With improvement of filtering initialization and model probability updating in IMM algorithm, multi-time model probability distribution based multiple model (MTMM) algorithm and model probability optimization based multiple model (MRMM) algorithm are built. On base of MSDA algorithm and MMSDA algorithm, IMM-MSDA algorithm and IMM-MMSDA algorithm are designed for multiple maneuvering target tracking. To decrease influence of glint noise in maneuvering target tracking, model set switching based tracking (MSGT) algorithm is put forward.3. Asynchronous track fusion algorithmFor asynchronous track fusion problem in maneuvering target tracking, optimal weight based multiple model (MMASTF) algorithm is proposed. With combination of asynchronous track fusion and out-of-sequence measurement processing, ATFOOSM algorithm is composed.