Research on the Intrusion Detection Model for Power Internet of Things Combining Deep Belief Network and BiLSTM

With the development of modern power systems, the integration of the Internet of Things (IoT) in power networks, namely the Power Internet of Things (PIoT), plays a crucial role in improving the efficiency and reliability of energy distribution. However, this advancement also brings significant cybersecurity challenges, which may lead to attacks on critical infrastructure. This paper proposes a deep learning-based intrusion detection model for the Power Internet of Things, which combines the architectures of the Deep Belief Network (DBN) and the Bidirectional Long Short-Term Memory network (BiLSTM). The model addresses the common issues of data complexity and high dimensionality in Power Internet of Things systems by utilizing DBN for data dimensionality reduction and feature extraction. The BiLSTM component captures the temporal dependencies in data streams, thereby enhancing the model’s ability to detect both known and novel intrusion patterns. Experimental results show that the DBN-BiLSTM model significantly improves the detection accuracy while maintaining real-time processing capabilities, which is essential for protecting IoT-driven power systems. This paper also explores further optimizations, such as reducing computational complexity through the CNN-BiLSTM combination, enhancing the model’s robustness and its ability to adapt to dynamic environments. This intrusion detection method provides a powerful tool for ensuring the stability and security of smart grids and contributes to the development of green and sustainable energy systems by mitigating cybersecurity risks in power systems. Keywords: Power Internet of Things, Intrusion Detection, Deep Learning, Deep Belief Network, BiLSTM, Cybersecurity, Smart Grid, Sustainable Energy