Intrusion detection method under group learning architecture fused with knowledge distillation

The invention relates to an intrusion detection method under a group learning architecture fusing knowledge distillation, and belongs to the field of mobile communication. The method comprises the following steps: constructing a knowledge distillation-based group learning architecture in a server with sufficient resources by using a digital twin network to train a teacher model; taking the unmanned aerial vehicles with rich resources as leader nodes, calculating the contribution degree of each unmanned aerial vehicle leader node, and taking the unmanned aerial vehicle with the highest contribution degree as a team leader node to be responsible for aggregation of global teacher models in the group learning network; in order to improve the quality of a teacher model, a joint optimization problem of reliability, time delay and energy consumption is modeled, and a digital twinborn assisted deep reinforcement learning algorithm is provided to solve the optimization problem; and training an intrusion detection mode