基于加权Logistic回归模型的森林火灾预测

Forest Fire Prediction Based on Weighted Logistic Regression Model

林火预警是保障超高压输电网安全的重要工作,是森林防火部门和电网公司深度关注的领域。研究以2007—2017年广东省超高压输电网线路途经地区的气象数据和林火监测的数据为基础,通过加权Logistic回归分析方法构建了广东省超高压输电线路区域森林火险预警模型,并用2017年实际林火发生数据对该模型进行检验。模型预测准确率达到92.6%,证明该模型具有良好的预测效果,反映了广东省区域森林火险等级与相关气象因子的密切关系。

Based on the meteorological data and forest fire records from 2007 to 2017, the dynamic weight of forest fire was obtained by cluster analysis, and the weighted Logistic regression model was established.Finally, the nonlinear regression equation between forest fire risk, meteorological factors and forest fire historical distribution factors in Guangdong province was obtained. The data of forest fire in 2017 were predicted by the equation. The accuracy rate is 92.6%. The availability of the model was verified. Reflecting the close relationship between the regional forest fire risk rating and related meteorological factors in Guangdong province. It can provide reference for the forest fire warning work of the ultra-high voltage transmission network in Guangdong province.