Abstract: The network flow planning method of LEO constellation is the key technology to improve the network performance of large constellation. Aiming at the problem that the existing network flow planning method of LEO constellation has a single means, which is difficult to take into account the global performance optimization and satellite autonomous flexibility, a flow network planning method of LEO constellation based on combination strategy is proposed. According to the service characteristics, the network flow is classified and mapped. Aiming at the QoS differences of six mapping services, the centralized, distributed and source end routing control strategies are adopted respectively; on this basis, the constellation network flow planning problem is abstracted to solve the maximum flow problem of multi-source and single sink networks and the maximum flow problem of multi-source and multi sink networks under bandwidth constraints, using load balanced routing and multi constraint QoS routing algorithms respectively; the constellation network flow planning method based on combination strategy and single strategy is simulated and analyzed. The results show that the combination strategy is about 20ms lower than the centralized transmission delay, the packet loss rate is about 9% lower, and the bandwidth utilization rate is about 42% higher than the distributed strategy. It can effectively improve the constellation network performance, and can better meet different scenarios and needs.