This paper proposes a layered detection response scheme to address the problem of unmanned aerial vehicle (UAV)self-organizing networks carrying important information but being susceptible to various attacks.Firstly,the end node misuse detection method is adopted to monitor the behavior of drones and report the results of rule validation. Subsequently,the ground station utilized deep neural network classification algorithms to further optimize the validation results.Provided new solutions to resist different types of network attacks such as interference,black holes,and gray holes.Finally,this article conducts simulation analysis and experimental comparison with the BRUIDS scheme and distributed detection scheme.The results show that compared to the other two schemes,the decrease rate of detection rate in our method remains within 10%,with a detection rate of over 93%.The false alarm rate has increased by 1.2%, which is almost the same as the distributed scheme in detecting attack latency.The communication overhead has been reduced by about 53.5 KBps,and there is a significant improvement in the detection of unknown attacks.