An Intrusion Detection Model of Incorporating Deep Residual Shrinking Networks for Power Internet of Things

With the rapid development of the Power Internet of Things (PIoT), the security of the power system is receiving increasing attention. Intrusion detection, as an important means of protecting the power system from malicious attacks, is crucial for ensuring the stable operation of the power grid. However, imbalanced data categories and insufficient ability of neural networks to handle noisy data limit the performance and reliability of detection systems. To address the aforementioned issues, this paper proposes a new intrusion detection model for power Internet of Things systems, which comprehensively utilizes Deep Residual Shrinkage Network (DRSN), SMOTE algorithm, and Generative Adversarial Network (GAN) to improve the effectiveness of data processing and feature extraction. The model uses SMOTE algorithm and GAN to generate minority class data and construct a balanced dataset. Using DRSN to denoise signals and enhance the ability of neural networks to extract features from noisy data. Extract features through Transformer and BiLSTM networks to capture long-term dependencies; Finally, use the Softmax classifier to obtain the classification results. Through testing on the NSL-KDD dataset and comparing it with other advanced algorithms, the results show that the proposed model outperforms other algorithms in terms of accuracy, precision, recall, and F1 score.