Low earth orbit satellite multi-beam subcarrier dynamic scheduling method based on deep reinforcement learning

The invention relates to the technical field of mobile communication and wireless communication, in particular to a low-orbit satellite multi-beam subcarrier dynamic scheduling method based on deep reinforcement learning, which comprises the following steps: setting an intelligent agent on a satellite, and updating the priority of a user by the intelligent agent according to a priority adjustment factor; judging whether the transmitting power for the queue is greater than the maximum transmitting power of the satellite after the user with the highest priority is added into the user queue, and if so, outputting the user queue; otherwise, adding the user with the highest priority into the user queue; calculating the number of subcarriers allocated to newly added users; after subcarrier distribution is completed, the throughput, the time delay and the packet loss rate of the current communication system are updated, and the users newly added into the user queue are deleted from the user set to be accessed; updat