To improve the efficiency of unmanned aerial vehicle(UAV) detection of ship exhaust emissions,  the path planning problem of UAV when both UAV and ship were moving simultaneously has been investigated. Aiming at  the real-time changes of the  ships position during the detection process, the "meeting" model was adopted to solve the encounter position of ships and UAV, and a UAV path optimization model targeting the minimum flight distance of moving ships was established to achieve collaborative detection of multiple UAVs accordingly. The two-stage algorithm by combining sequence insertion algorithm  and a genetic algorithm based on two ship sequence decomposition strategies, DroneByDrone and ShipByShip were designed to achieve UAV path optimization in scenarios of different scales. Numerical experiments show that a path optimization method that taking into account of both UAV flight and ship navigation can effectively improve the detection efficiency of moving ships. Both algorithms can successfully solve the model, but compared to the two-stage algorithm, the genetic algorithm can shorten the solving time by 65.12%, meeting the requirement of solving timeliness. The  flight distance  of UAVs is significantly sensitive to the number of UAVs, in the same dataset, dispatching two UAVs can optimize the flight distance by 25% compared to three UAVs, but the detection time will be correspondingly delayed by 8%.By comprehensively optimizing the position of UAV base stations, the number and speed of UAVs based on  experimental scenarios, the  flight distance of UAVs can be effectively shortened, and the detection efficiency can be improved.