改进型灰色神经网络在火灾预测中的应用研究

为提高火灾事故的预测精度,降低火灾损失,探讨并修正传统GM(1,1)预测模型关于背景值构造的缺陷,将改进后的灰色模型同BP神经网络模型融合,提出改进型灰色神经网络火灾预测模型。依据我国1997—2009年火灾事故统计数据,分别选用改进型GM(1,1)和改进型灰色神经网络模型对1997—2007年火灾发生起数进行拟合仿真,得到2008—2009年火灾起数预测结果。结果表明:该模型在避免GM(1,1)关于背景值构造缺陷的基础上,兼具灰色系统与神经网络的优点,既体现火灾复杂的灰色系统行为,又能自适应调整学习速率,与单一GM(1,1)相比,该模型的预测结果精度更高。

Research on Application of Improved Gray Neural Network in Fire Forecasting

In order to improve the prediction accuracy of fire and reduce fire loss,the structural defects concerning background values in the traditional GM(1,1) model were discussed and tackled.The improved gray model and the BP neural network model were integrated to present the improved gray neural network prediction model for fire accidents.According to the statistical data of 1997-2009 fire accidents,the improved GM(1,1) model and the improved gray neural network model were respectively used for fitting simulation of the 1997-2007 fire disaster occurrence number,the prediction results of 2008-2009 fire disaster occurrence number were obtained.The results show that this new model avoids the structural defects concerning back ground values in the GM(1,1) model,and has the advantages of both the gray system and the neural network.Moreover,it not only embodies the complex gray system behavior of fire and could adaptively adjust the learning rate.Compared with the single GM(1,1) model,the predicted result of this new model owns the higher precision.