State Monitoring of Power Distribution Systems Based on ZigBee Technology and Graph Attention Networks

Due to the natural topology of the power grid, current technologies fail to quickly locate abnormal positions and perform fault classification diagnosis when anomalies occur in power distribution systems. Therefore, a state monitoring model for power distribution network systems is proposed in this study. This study first combined ZigBee technology and multi-core graph attention network to construct the state location transfer network of the power distribution network system, and on this basis, applied the Internet of Things and recurrent neural network to achieve real-time data collection and processing. Subsequently, the collected data are input into the graph neural network and the temporal convolutional network. Through feature extraction and pattern recognition, anomaly location and classification diagnosis are achieved. Experimental results show that when the number of nodes reaches 90, the end-to-end delay of the model is 0.91ms, the routing deployment cost is 1700 USD, the packet delivery rate remains above 91.2%, the accuracy of anomaly localization exceeds 94.8%, and the response time does not exceed 5.1ms. All these results outperform the compared networks. In comprehensive performance verification experiments, the model achieves an average diagnosis accuracy of 97.8% for five types of anomalies. The errors in localization accuracy and classification accuracy are 0.031 and 0.017, respectively, both significantly better than those of the compared models. The proposed state monitoring method for distribution network systems in this study enhances the real-time performance of data transmission and enables precise positioning of abnormal node states, providing a strong guarantee for the stable operation of the power distribution network system. This study provides ideas for the state monitoring of intelligent power distribution network systems.