摘要
为了智能化预测森林火情,在分析火情传感器的信息参量的基础上,提出了模糊隶属函数与神经网络相融合的预测方法。利用欧洲标准试验火TF1模型的运算分析表明,该方法能有效降低概率在0.5附近的森林火情误报率;进一步引入干扰信息参量后预测时间虽有短暂的延迟,但仍能比较准确地预测火情并输出其概率特征。该文提出的模糊神经网络研究方法对复杂度较高的森林火情传感网络及其预测系统具有较强的实用价值。
Intelligent processing for forest fire forecastings requires information from a range of fire sensors. A fuzzy membership function and neural network suitable for forecasting were developed for a fuzzy-neural network for fire forecasting. Experiments with the TF1 model of the European standard test fire show that this fuzzy-neural network effectively reduces the false alarm rate for forest fires near a probability of 0.5 and reasonably forecasts forest fire with a probability feature output even with a delay due to interference information. This fuzzy-neural network based method may be applicable to complex fire sensor networks for forecasting.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2010年第8期1302-1306,共5页
Journal of Tsinghua University(Science and Technology)
基金
国家"九七三"重点基础研究发展计划项目(2007CB310601)
关键词
森林火情预测
模糊神经网络
火灾概率
forest fire forecasting
fuzzy-neural network
fire probability
