Research on the Application of Cloud Based on Data Fusion and Visualization in Line Monitoring

This paper propose a topology aware graph attention cloud data fusion (TG-ACDF) method to address the issues of low accuracy and lack of spatial correlation modeling in multi-source heterogeneous data fusion for power line monitoring. TG-ACDF models the line monitoring network as a graph structure, extracts topological features through graph guided feature augmentation (TFA), and deeply integrates them with physical features to enhance feature representation capabilities. Then, self evolving spatial temporal attention (SESA) simultaneously learns spatial correlations and temporal evolution laws. The experimental results show that our TG-ACDFRMSE is reduced to 1.42, F1 Score reaches 0.91, and exhibits good robustness in noise interference and data loss scenarios. The cloud based distributed architecture based on graph segmentation supports real-time monitoring of large-scale power line networks, providing effective technical support for smart grid operation and maintenance. TG-ACDF significantly outperforms existing methods in abnormal fault warning and feature fusion accuracy. F1 Score reached 0.91 and RMSE decreased to 1.42. The cloud based distributed architecture based on graph segmentation achieves an acceleration ratio of 7.13 times at a scale of 100 nodes, with inference time controlled within 3.42 seconds, meeting the monitoring requirements of large-scale line networks.