Hypersonic Vehicle Tracking Algorithm Based on Virtual Radar Constructed by Artificial Intelligence

Hypersonic vehicles have complex maneuvering modes and changeable flight trajectories. Therefore, the prior prediction model established for a single flight situation has the problems of low fault tolerance and high modeling cost. To solve this problem, this paper designs a multi-source information fusion filter based on virtual radar constructed by artificial intelligence, which uses a data-driven way to obtain flight information of non-cooperative hypersonic targets. The core principle is that one of the sub-filters uses the support vector regression machine to learn the historical trajectory data online, and provides a predicted value after each radar sampling period, which contains the information of the historical flight trajectory of the hypersonic vehicle. The predicted value is set as a virtual radar sampling result and introduced into the prediction model for filtering and correction, while the input of another sub-filter is the real radar sampling data, and finally the outputs of the two are fused. The simulation results show that the algorithm has stronger fault tolerance and accuracy when tracking hypersonic vehicles.