To deal with the problem of cooperation of multiple heterogeneous UAVs for target searching and attacking in unknown environment, and with consideration of the resources of the UAVs and the targets, a method of coalition formation was used to complete the task.A multi-objective optimization model was established, and a parallel Non-dominated Sorting Genetic Algorithm(NSGA-II) was proposed to solve the problem. A simulation was carried out and an analysis was made to the running speed of the algorithm, which was compared with the traditional methods. The simulation results verified the rationality of the model, and showed that the parallel NSGA-II has a strong real-time performance and improves the efficiency of task performing.