Online Attack Intent Recognition for Aerial Vehicles Based on Interacting Multiple Models

To address the complexity challenges in combat intention recognition for aerial vehicles’ swarm cooperative operations, this paper proposes a guidance-characteristic-integrated recognition method. A dynamic self-adaptive hybrid algorithm framework combining Cubature Kalman Filter (CKF) and Interacting Multiple Model (IMM) is developed to effectively overcome the limitations of conventional approaches in motion feature extraction. Specially, each possible attack intent of the vehicle is modeled as interception motions filter using the Proportional Navigation Guidance (PNG) law, and the IMM algorithm is utilized to achieve multi-model interaction and dynamic coupling of multiple models. Experimental verification demonstrates that the proposed method achieves accurate recognition of clustered targets’ initial attack intentions while exhibiting adaptive capabilities in dynamic scenarios involving target switching between attack objectives, thereby enabling real-time identification of attack intent. Compared with conventional generalized motion models, the identification speed is improved by over 40%, significantly enhancing the real-time performance of target recognition and the timeliness of countermeasure decision-making in adversarial engagements.