The assembly station is the basic management unit of the aircraft assembly line. Due to the complicated process of aircraft assembly and a large number of random disturbances, its managers need to frequently optimize the material configuration of the aircraft being processed. To this end, an optimization method based on the gated recurrent unit (GRU) neural network and genetic algorithm was proposed. In order to overcome the limitation of discrete event simulation in terms of efficiency, a simulation agent model of material configuration evaluation based on GRU neural network was constructed by taking the simulation historical data as the learning sample. The model took the material configuration as the input, and took the estimated completion time and the average residence time of key materials as the output. The simulation agent model was combined with the genetic algorithm as the objective function evaluation model to realize the global optimization of the material configuration. The simulation verification results show that the simulation agent model based on GRU neural network can accurately and efficiently evaluate the material configuration, and the output optimization configuration can effectively shorten the estimated completion time and average residence time.