Joint Probability Evaluation Model and Risk Visualization Method of Line Outage Risk Caused by Wildfires Based on PSO - RF

Wildfire disasters can easily cause overhead transmission lines to trip, threatening the stable operation of the power grid. In order to accurately predict the risk probability of transmission line trip induced by wildfires, this paper considers the influence of historical meteorological data and environmental vegetation characteristics on the probability of wildfire tripping, and proposes a joint probability evaluation model of line outage risk caused by wildfires based on Particle Swarm algorithm to optimize (PSO) Random Forest (RF). Firstly, the dataset of line outage caused by wildfires was constructed based on the historical data measured by North China Power Grid, which included meteorological spatiotemporal data and line environmental vegetation characteristics, with a total of 489 samples. Secondly, an RF-based joint probability evaluation model for wildfire tripping of transmission lines was proposed, and the PSO algorithm was used to optimize the RF hyperparameters to further improve the prediction accuracy of the model. The above feature datasets are used to train and test the performance of the proposed model, and the logistic regression, linear discriminant, neural network and gradient boosting models are compared in turn, and the results show that the performance of the proposed model is the best. Finally, the random forest algorithm was used to predict the probability of wildfires on transmission lines, and the natural discontinuity method was used to classify the fire risk at five levels, and the fire risk distribution map was drawn. Based on the wildfire occurrence model and the wildfire-induced tripping model, the joint probability evaluation model of wildfire tripping of transmission lines is obtained, with an accuracy of 95.71% and a confidence of 90.99
^%
, which can provide some theoretical guidance for disaster prevention and mitigation of power grid.