灰色Elman神经网络在火灾事故预测中的应用研究

针对我国火灾事故现状,结合灰色理论和神经网络的特点,提出灰色Elman神经网络火灾事故预测模型。依据我国1998—2007年火灾事故统计数据,分别选用GM(1,1)模型和灰色Elman神经网络模型对1998—2005年数据(火灾起数)进行拟合,对2006年、2007年数据进行(火灾起数)预测。结果表明:灰色Elman神经网络火灾事故预测模型符合火灾事故的特点;有效地解决了传统灰色预测模型在火灾事故预测中误差大稳定性差的缺陷,提高了预测精度;可对火灾事故进行预测与分析,为消防安全管理提供依据,以最大限度地减少火灾事故的发生。

The Application Study of Gray Elman Neural Networks to Fire Accident Prediction

A gray Elman network model for fire accident prediction based on gray theory and neural networks is put forward.Then,the fire data(fire frequency)from 1998 to 2005 are fitted respectively by GM(1,1)and the gray Elman network model,and the number of fire accidents in 2006 and 2007 is predicted.The results show the gray Elman network model is suitable for the fire accident prediction;it can effectively solve the problems of big error and weak stability existing in the traditional gray prediction model.