With the development of 5G technology and the research of 6G technology， the LEO satellite network will play an increasingly important role in the future integrated air-space-ground network. However， the routing strategy as the core technology of the network still faces some challenges， such as high dynamic changes in satellite network topology， frequent switching of paths， limited computing and processing capacity of nodes， and unbalanced load and traffic distribution. Aiming at the problem of unbalanced load in the area covered by the satellite and related to time factors， and the multi-service QoS requirements of users on the satellite， this paper designs a space-time-level-based service classification load balancing routing algorithm. The algorithm considers the relationship between the distribution of satellite network traffic and the space-time level. The satellite node needs to solve its space-time level according to the current moment and the ground area it covers. In the next hop real-time adjustment stage of the routing algorithm， the space-time level is added as one of the routing conditions. At the same time， the calculation method of the threshold is improved， and different link weights are considered when routing to meet the multi-service QoS requirements. The simulation results show that， compared with the DSP algorithm， the TLR algorithm and RMLBR algorithm， although the algorithm in this paper is not as good as the TLR and RMLBR algorithm in terms of total throughput， the delay and ISL utilization have been improved， which effectively reduces the average end-to-end delay， better meets the QoS requirements of different services， and balances the network load.