摘要
为了提高矿井火灾探测器对环境的适应力和抗干扰能力,采用逼近能力、分类能力和学习速度等方面优于BP网络的径向基函数神经网络,在MATLAB环境下构建火灾探测仿真模型,以温度、烟雾浓度、CO气体浓度作为输入,进行多信息数据融合,达到矿井火灾探测目的。仿真结果表明,该方法对明火、阴燃火和无火概率的识别误差均小于5%,可大幅降低火灾报警的漏报和误报率。模糊系统和神经网络相结合的手段,能有效监测矿井火灾的产生,对于智能火灾报警系统研究具有参考价值。
In order to improve the environment suitability and the anti interferences capacity of the mine fire disaster detector,the radial basis function neural network with the approximation capacity,classification capacity and learning speed better than the BP network was applied to establish the mine fire disaster detection simulation model under the MATLAB environment.The temperature,smoke density and CO density was applied to the input for the multi information data integration to reach the target of the mine fire disaster detection.The simulation results showed that the identification probability error of the open fire,the shade fire and no fire by the method would be all less than 5% and the method could highly reduce the missed detection and incorrect detection rate of the fire disaster early warning.The means combined with the fuzzy system and the neural network could effectively monitor and measure the mine fire disaster to be occurred and would have the reference value to the study on the intelligent fire disaster warning system.
出处
《煤炭科学技术》
CAS
北大核心
2011年第2期65-68,共4页
Coal Science and Technology
基金
2008自主课题资助项目(SKLCRSM08B12)
"十一五"国家科技支撑计划资助项目(2007BAK22B04)
关键词
径向基函数
神经网络
MATLAB
智能火灾探测
radial base function
neural network
MATLAB
intelligent fire disaster detection
