In view of the problem of collaborative real-time trajectory planning during the execution of ground combat missions by multiple unmanned combat aerial vehicles (UCAV), a real-time 3D trajectory planning method based on the sine cosine optimization algorithm with self-learning strategy and Lévy flight (SCASL) for multi-UCAV cooperatively attacking multi-targets was proposed to improve the real-time performance and operability of UCAV collaborative real-time trajectory planning. Firstly, the model of 3D mission space was constructed, and the constraints of flight speed, flight altitude, relative distance, and threat avoidance were designed according to the performance of UCAV. Secondly, the decision variables of collaborative real-time trajectory planning were designed based on the three-degree-of-freedom kinematics and dynamic particle model of UCAV. Finally, the objective function was constructed by transforming the collaborative real-time trajectory planning problem into an optimization problem according to the tactical principle of UCAV operations. The proposed SCASL was applied to solve the model, and the virus search algorithm (VCS) was adopted for comparison. Simulation results show that under the same conditions, the results obtained by VCS met the requirement of collaboration but not real-time, while the results obtained by SCASL met the requirements of both real-time and collaboration, which verifies the validity and superiority of the SCASL-based multi-UCAV collaborative real-time 3D trajectory planning method proposed in this paper.