This paper studies the multi-UAV task allocation problem in large-scale scenarios, which involves the constraints of task timing and heterogeneous resources. However, existing algorithms for this problem tend to ignore the balance of resources and suffer from low solution efficiency and insufficient performance, thus limiting potential applications of UAVs in complex environments. To solve these problems, this paper first cluster the targets based on the k-means algorithm, and then performs group matching between mission targets and UAV groups based on inter programming. Next, the traditional distributed auction algorithm is improved to assign targets to UAVs with heterogeneous resources according to task timing constraints. The proposed scheme improves the accuracy and efficiency of task allocation, reduces the communication between UAVs, and enhances cooperative operation capability of UAV swarm. Finally, the feasibility and effectiveness of the scheme are verified by simulation experiments.