High accuracy wind-induced response prediction of transmission tower based on graph neural hybrid network

Response prediction is critical for structural safety assessment and disaster resilience for transmission towers. Due to the inherent characteristics of transmission towers, such as their towering height, heavy load, and strong flexibility, they are particularly sensitive to wind load excitations. In this paper, we propose a response prediction model based on a Graph Attention Neural Network (GAT-ResLSTM) for transmission towers under wind load conditions, which transforms the transmission tower into a graph data structure and deeply explores the spatial and temporal characteristics of different nodes, and their dependencies on each other to realize the high-precision temporal response prediction of the transmission tower. To verify the validity of the model, the transmission tower under wind conditions is taken as an example for numerical validation, and the prediction performance of the model on the field monitoring data in windy conditions of the transmission tower is also studied. The results show that the proposed model has good prediction performance in transmission tower response prediction, which is better than the traditional time series prediction model, and the error in monitoring data prediction is less than 4 %.