Intelligent aerial combat involving unmanned aerial vehicles holds transformative significance for future warfare. Significant progress has been made in recent years regarding intelligent decision-making and control for tracking maneuvering targets. However, challenges persist in tracking fighter jets performing post-stall maneuvers that rapidly alter flight position and state, resulting in low tracking accuracy or even target loss. This hinders the application of missile guidance systems for precision strikes. To effectively address this issue, a guidance system design method based on intelligent decision-making for target motion is proposed. This method considers the complete process of several typical post-stall maneuvers, decomposing them into four fundamental motion modes: low-angle-of-attack flight, rapid pitch-up, rotation around the velocity vector, and rapid pitch-down. Aiming to minimize motion modeling errors, it employs a long short-term memory neural network to predict the probability distribution of the target being in each of these four motion modes. Furthermore, control strategies adapted to the characteristics of each motion mode are adopted to enhance tracking accuracy. Simulation results demonstrate that the proposed method enhances tracking robustness and accuracy while ensuring no loss of post-stall maneuvering targets.