In the scenario of game confrontation, S-maneuver and barrel roll maneuver are generally used to avoid missile strikes against airborne targets. When the target is maneuvering, the guided missile's trajectory bends, which can easily cause energy loss,and poor missile-target rendezvous can decrease the effectiveness of combat coordination，then the realization of high-mobility target trajectory prediction is of great significance to the realization of predictive guidance and thus the improvement of ballistic performance. To solve the problems of complex target maneuvering patterns and difficult modeling, an adaptive prediction method based on target maneuvering trajectories was proposed. To meet the massive data requirements of intelligent training, design of generating data sets and test sets which mixed modeling using experimental data、 digital and simulation were proposed. and CNN-LSTM neural network was designed to improve the adaptability and generalization ability of the network model, which significantly improving the recognition accuracy and precision. The experimental results demonstrate that the target trajectory prediction method can adapt to various combat scenarios such as target maneuvering styles and parameter uncertainties.