Mountain fires pose a threat to the safe operation of transmission lines.Firstly,the historical pattern of mountain fire incidents in Hunan province from 2018 to 2020 is reviewed.On this basis,a database of mountain fire incidents is constructed.One-hot encoding technique is introduced to numerically process textual features such as the vegetation type along the transmission lines.Then,XGBoost technology is utilized to build a mountain fire risk assessment model for transmission lines.Aiming at imbalanced samples between forest fire incidents and normal operations,based on a cost-sensitive mechanism,a weighted objective function is presented to mitigate the problem of overlooking mountain fires caused by sample imbalances.Finally,the proposed mountain fire risk assessment model for transmission lines is tested using the forest fire incidents in Yongzhou City,Hunan Province,from 2020 to 2021 to validate its effectiveness.
关键词
山火风险评估/样本不平衡/代价敏感机制/独热编码/极端梯度提升树
Key words
forest fire risk assessment/sample imbalance/cost-sensitive mechanism/one-hot encoding/XGBoost
维普
万方数据