Research On The Variable Structure Multiple-model Maneuvering Target Tracking Algorithm

Technology of maneuvering target tracking is an important research subject all the time in the information fusion system, and it has been widely used in civil and military field. With the target of mobility being stronger and stronger, and accuracy demand of the maneuvering target tracking is increasingly high, it is particularly important to realize the maneuvering target tracking steadily, accurately and reliably. Perfecting and improving the existing target tracking technology has its theoretical and practical significance. Multiple-model algorithms provide a good solution for the problem of maneuvering target tracking. With the constant development and progress of the multiple-model algorithms, the variable structure multiple-model algorithm has gradually become the mainstream of research. More and more scholars propose different variable structure multiple-model algorithms, which largely improves the target tracking precision and real-time performance. This paper presents three new variable-structure multiple-model algorithms based on the existing variable-structure multiple-model algorithm. The main content is organized as follows:1. Variable structure multiple-model algorithm based on likely model set and best model augmentation(LMS-BMA), the main idea of algorithm is that first the basic model-set utilize the model set self-adaptation in likely model set algorithm to delete the unlikely models in itself and to get the effective model set of the current moment, and then use the KL criterion of best model augmentation algorithm(BMA) to select the model set current moment’s best model from the predesigned candidate model set, finally use the model set sequence conditioned estimation based on the effective model set and the best model to obtain the overall of maneuvering targets. The simulation result of comparison with the existing algorithms verifies that the proposed algorithm has higher cost-effectiveness.2. Variable structure multiple-model algorithm based on expected-best model augmentation(EBMA), the main idea of the algorithm is that first the basic model set use expected-mode augmentation algorithm to obtain the expected model of current moment, then use Euclidean distance between the expected model and the candidate model set to select the model that nearest distance of the candidate model to expectedmodel as the expected-best model of the current moment, finally use the model set sequence conditioned estimation based on basic model set and the expected-best model to obtain the overall estimate of maneuvering targets. The simulation result of comparison with the existing algorithms verifies that the proposed algorithm has higher cost-effectiveness.3. Variable structure multiple-model algorithm based on expected-model and trade-off model augmentation, the main idea of the algorithm is that first the basic model set use expected-mode augmentation algorithm to obtain the expected model of current moment, then obtain the trade-off model based on the intersection of straight line made up of expected model and the nearest basic model and line between two nearest basic models, finally use the model set sequence conditioned estimation based on basic model set and the trade-off model to obtain the overall estimate of maneuvering targets. The simulation result of comparison with the existing algorithms shows that the proposed algorithm has great improvement in tracking performance.