火灾发生的周期性规律及其在火灾非线性组合预测模型中应用

研究火灾发生规律及发展趋势,具有实用价值.分析历年中国火灾统计数据,发现中国火灾的发生规律同时具有增长趋势性和周期波动性特征:①中国火灾从90年代开始大幅度增加,在2002年达到最大,然后逐年缓慢下降;②每年12个月呈正弦函数波动,周期为12,1、2月发生起数最大,8、9月最小.借助于M ATLAB软件,根据2000-2006年中国火灾统计数据,建立火灾的月发生起数的非线性周期性组合预测模型,预测值与实际值的平均相对误差小于0.07.研究结论为消防研究、消防部门决策提供科学依据.

Analysis of Fire Periodicity and its Application to Non-linear Combinatorial Forecasting Model for Fire

It is valuable to study fire regularity and trend. Based on statistics of fire in China in the past years, the characteristics of developing trend and periodical fluctuation are found in fire regularity in China, that is: 1) the number of fire in China is increasing rapidly since the 1990s, and to its largest in 2002 before its slow descent; 2) the number of fire occurrence fluctuates in a way of Sine Function in a period of 12 months with the largest in January or February and smallest in August or September. Based on MATLAB and statistics of fire in China from 2000 to 2006, non-linear periodicity combinatorial forecasting model for fire occurrence by the month is established, only to find the average relative error between predicted and observed value is within 0. 07. The result can provide scientific guidance for fire study and fire department as well.