Abstract:

In order to solve the problems of the high maneuvering target tracking algorithm with fixed structure multiple models, which are difficult to meet the real-time requirements of the system, such as large computation amount and long computation time, a variable structure multiple model (VSMM) algorithm based on the target maneuvering pattern recognition is proposed. This algorithm can select the model set that matches the target motion state, and has the advantages of time-varying and self-adaptive. By using a small number of motion models to form a model group, and then combining different motion models to form different model groups, the proposed algrithm replaces the multiple models in the original fixed structure model with the model group, and selects whether to replace the model group according to the detected target motion pattern, so as to obtain a more accurate suboptimal solution with a small number of operations. The simulation results show that the algorithm can not only reduce the computational burden effectively, but also make the model set and the motion state of the target better matched, so as to improve the tracking accuracy of the target.

Key words: maneuvering target tracking, variable structure multiple model(VSMM), pattern recognition, adaptive