With the rapid development and popularization of low orbit satellite networks, their importance in global communication coverage and data transmission continues to be highlighted. However, due to the particularity of LEO(Low Earth Orbit) satellite networks, such as highspeed operation, limited channel capacity, and large latency, poses challenges to network load balancing and routing scheduling. The current routing algorithms for low orbit satellite networks mostly use traditional methods such as Dijkstra's shortest path, which lack real-time adaptability and intelligence to the high mobility of the network. Therefore, based on the deep reinforcement learning intelligent routing algorithm of DQN(Deep Q Network) has become a new approach to solve this problem. By combining the multidimensional information of LEO satellite networks and the advantages of DQN, this study proposes a new load balancing intelligent routing algorithm that comprehensively considers inter satellite link connectivity and link load conditions, and jointly optimizes multiple objectives. The aim is to improve network performance, reduce latency, and provide technical support and solutions for the future development of low orbit satellite networks.