Since most traditional resource allocation methods only focus on a single problem, which are difficult to expand and time-consuming to solve, in this paper, we propose a novel resource allocation method with the IMPALA algorithm. Based on the deep reinforcement learning formulation, the states, actions and rewards are designed, and then the IMPALA is introduced to solve the concrete problem. Our method can obtain the automatic planning of the resource allocation with powerful decision-making ability. Experimental results based on simulation demonstrate that the proposed method can not only retain the similar performance to the traditional methods like Particle Swarm Optimization (PSO) but greatly improve the computing efficiency.