基于知识蒸馏与不确定性估计的分布式异常检测OA

A distributed anomaly detection scheme based on knowledge distillation and uncertainty estimation is proposed to meet the real-time and accurate requirements of power Internet of Things systems.The scheme first constructs a high-precision teacher model of anomaly detection based on the multi-scale spatiotemporal residual network and Gate Recurrent Unit,enhancing spatiotemporal data features'deep representation and generalization ability.Then,utilizing knowledge distillation techniques to generate lightweight and efficient student models to meet the requirements of distributed deployment.Finally,utilizing uncertainty estimation to optimize the anomaly detection results of distributed nodes further enhances the detection performance of the model.The experimental results show that the anomaly detection accuracy of the proposed system can reach 95%,the time efficiency of model training is improved by about 60%,and with better generalizability.