摘要
通过数据拟合认为2007-2016年济南市各季度及24 h时段的过火面积-火灾频率符合幂律分布,并计算得到拟合参数。将ARIMA模型的差分方程形式与传递形式结合,同时引入记忆函数进行优化,构建实时自修正预测模型,研究城市火灾的时间变化规律。研究结果认为,新构建的模型能够更好地预测城市火灾。
By data fitting, it was believed that the quarterly and 24 h period burned fire-fire frequency of Jinan city in 2007-2016 fits power law distribution, and the fitting parameters were calculated.By combing the difference equation form and transfer form of MAIMA model, and optimization by memory function, real-time selfcorrecting model was built to study the law of city fire time variation. The study showed that the new model can forecast the city fire well.
作者
吴立志
吴彬航
WU Li-zhi;WU Bin-hang(China People's Police University,Hebei Langfang 065000,China;Jinhua Fire and Rescue Detachment,Zhejiang Jinhua 321000,China)
出处
《消防科学与技术》
CAS
北大核心
2019年第9期1318-1322,共5页
Fire Science and Technology
关键词
城市火灾
时间特征
幂律分布
预测值自修正模型
urban fire
time characteristics
power law distribution
predictive value self-correction model