The trajectory planning of unmanned ships is an important part of unmanned ships autonomous navigation. A trajectory planning method for unmanned ship dynamic collision avoidance was proposed based on improved PRM algorithm and event triggering mechanism. Firstly, PRM and A* were combined to obtain the PRM-A* algorithm, and a trajectory planning model for unmanned ships was established based on the PRM-A* algorithm. A grid map was established based on the current environmental situation, the PRM-A* algorithm was used to plan the collision avoidance path of unmanned ships, and an S-T grid map of the collision avoidance path was established. The PRM-A* algorithm was used to plan the speed of unmanned ships on the S-T grid map, thus obtaining the collision avoidance trajectory of unmanned ships at the current time. Secondly, a dynamic collision avoidance model was established for unmanned ships based on the event triggering mechanism. An unmanned ship collision risk assessment model was established based on the TCPA and DCPA of unmanned ships, based on which event triggering condition was set. When the event was triggered, the unmanned ship collision avoidance trajectory was planned based on the current time window environmental situation. Finally, simulation experiments were conducted on the trajectory planning of unmanned ships in open and restricted waters. The results show that the algorithm can effectively make an unmanned ship avoid static and moving obstacles in waters, save computational resources, and improve computational performance.