In this research, an important topic of cooperative search for multi-dynamic targets in an unknown marine environment by unmanned aerial vehicle (UAV) is studied based on a novel multi-bee colony elite-learning algorithm. Firstly, a specialized searching model is established which includes the UAV dynamic, the sensor model, the target probability and environmental certainty with different flight altitude. Then, a new search strategy, which consists of rough search and precise search is proposed by considering the multi-objective cost function with the dynamic changing of the flight altitude. In order to solve the cost function, an improved multi-bee algorithm based on elite learning is designed, which overcomes the shortcomings of the poor adaptability and slow solution of the standard ABC algorithm, and ensures strong adaptability under different search. Finally, the improved algorithm and search strategy were simulated in different scenarios, and the results verify the effectiveness of the proposed method.