This article investigates the collaborative tasks and resource allocation of multiple unmanned aerial vehicles(UAVs). For multi-target collaborative attack tasks, this article optimizes the resource allocation of ammunition carried by UAVs based on factors such as target value, ammunition limit, UAV bomb load and range, in order to maximize the benefits of attack tasks. Firstly, the task and resource allocation problem was modeled, and an allocation algorithm based on probabilistic clustering is proposed, in which the agent's action strategy is assigned a probability value and updated to optimize individual and overall benefits. Secondly, the conflict resolution method and the optimal solution expansion strategy based on neighborhood sampling are introduced to solve problems in actual operation. The simulation results show that the proposed algorithm effectively solves the problem of task resource allocation, and can meet different solution efficiency and solution quality requirements according to the adjustment of the neighborhood radius and the number of cycles.