Aiming at the problem of target tracking of the armed helicopter fire attack, a method of target state estimation based on improved cubature Kalman filter (CKF) and adaptive multi-model interactive algorithm (IMM) is proposed. Considering the model error and measurement error of nonlinear dynamic system, the conventional CKF algorithm is modified based on robust M estimation theory reduce the influence of errors. Secondly, adaptive adjustment factors are introduced to deal with dynamic model errors and improve estimated accuracy. Finally, aiming at the problem that Markov matrix in conventional IMM algorithm remains unchanged in the tracking process, the Markov matrix elements are updated adaptively by using the model likelihood function value and the decision window to improve the model switching speed and matching degree. The simulation demonstrates that the enhanced algorithm effectively mitigates the impact of measurement noise and model error, significantly enhancing the accuracy and stability of 3D maneuvering target state estimation.