Risk Assessment of Transmission Tower in Typhoon Based on Spatial Multi-source Heterogeneous Data

Risk assessment and visualization of power system under typhoon disasters has scientific significance and engineering application value for disaster prevention and mitigation of power systems. In order to predict high-risk areas and optimize the emergency material allocation and risk-based dispatch of power flow, the data layer, knowledge extraction layer and visualization layer are used to construct the risk assessment system for power transmission towers under typhoon disasters. Firstly, based on equipment operation information, meteorological information and geographic information, a spatial multi-source heterogeneous information database is built. Then, based on parameter optimization, six machine learning algorithms are used to establish intelligent models for tower damage risk prediction, and a relative optimal model is selected through index comparison. At the same time, a combined model based on goodness of fit method with unequal weight is proposed. The tower damage risk in a Chinese coastal city under the typhoon "Mujigae" is assessed and visualized with dimension of 1 km×1 km. The relative optimal model is compared with the combined model in detail. The results show that both relative optimal model and combined model can identify the most severely damaged area, but the combined model has better prediction with the same risk threshold, which verifies the feasibility and rationality of the proposed method. Finally, the model universality and the influence of sample magnitude on prediction effect are analyzed. © 2020 Automation of Electric Power Systems Press.