Abstract: For high dynamic within visual range air combat of unmanned combat aerial vehicle (UCAV), this paper addresses tactical-maneuver trajectory optimization and trajectory tracking control. By jointly considering air combat situational metrics and maneuver energy, a lightweight reinforcement learning particle swarm trajectory optimizer and a total energy linearized tracking control framework are developed for four representative tactical maneuvers: Loop, Immelmann, High Yo-Yo, and Barrel Roll. First, an angle-of-attack (AOA) coupled thrust model and a load factor envelope model incorporating maneuvering energy information are established to support maneuver optimization and total energy control design under propulsion–airframe coupling. Then, using the rates of change of AOA, throttle, and velocity roll angle as command primitives, the segmented command models are constructed for the four maneuvers. An improved particle swarm optimization algorithm is then proposed by integrating a Q-learning-driven particle learning paradigm and a K-nearest neighbor differential-evolution mechanism, enabling maneuver trajectory optimization with respect to relative geometry, attitude, load factor, and thrust work-related metrics. Thereafter, the optimized trajectory commands are converted into total energy rate and energy-allocation rate commands. A total energy linearized tracking law is derived via linearization of the augmented total energy dynamics and desired time-domain response-based gain assignment. An adaptive correction factor based on online aerodynamic identification is further introduced to compensate for aerodynamic perturbations. Finally, numerical simulations on the four tactical maneuvers validate the proposed intelligent maneuver optimizer and total energy linearized controller. Results demonstrate reduced maneuver energy loss and optimization time, and significantly improved tracking accuracy and robustness under propulsion–airframe coupling and aerodynamic perturbations.

Key words: tactical maneuvers optimization, reinforcement learning, evolutionary particle swarm optimization, maneuvering energy, total energy control, AOA/thrust coupling