An improved adaptive IMM-CKF method for tracking hypersonic glide vehicles via space-based radars

This paper proposes an improved adaptive IMM method for tracking glide vehicles to address the limitations of traditional algorithms, including fixed transition probability matrices, low model selection confidence and large tracking errors. Within the IMM framework, a constant acceleration model, a J2 perturbation model, and a current statistical model were integrated, and the cubature Kalman filter algorithm was employed for tracking hypersonic glide vehicles. To adaptively update the transition probability matrix, three mechanisms are introduced: (1) monitoring changes in model probabilities to detect trends in model dynamics; (2) analyzing variations in model likelihood functions to identify model transition phases; (3) employing a decision window to determine model stabilization periods. Simulation show that the proposed method significantly outperforms classical algorithms, achieving increase in position accuracy by 14.5 and 11.1%, and in velocity accuracy by 19.1 and 8.7%, compared with CKF and IMM-CKF, respectively. Furthermore, it reduces model distribution entropy, enhancing model selection performance, and providing support for space-based radar target tracking applications.