基于神经网络的电缆火灾致灾因素分析建模与仿真

Modeling and simulation of cable fire factor analysis based on neural network

针对电缆火灾致灾定性、定量因素并存,且各因素存在时序特征与非线性关联的问题,提出了基于层次分析法与神经网络的电缆火灾致灾因素分析模型,并提高了因素分析的准确性和鲁棒性。该模型采用层次分析法对致灾因素进行逐级分析,建立了神经网络模型,对各定性、定量因素以及其特征值进行转换,并将其作为神经网络的输入进行模型训练,从而建立致灾因素与火灾之间的非线性时序模型。为了验证模型的有效性,基于真实数据对模型进行仿真实验。结果表明,采用神经网络对各因素的时序性和非线性完成建模后,可有效提高各因素与火灾预警的关联准确性。

In view of the coexistence of qualitative and quantitative factors of cable fire disaster,and the time-series characteristics and nonlinear correlation of each factor,proposes a cable fire factor analysis model based on AHP and neural network,which improves the accuracy and robustness of factor analysis.Based on the analytic hierarchy process,the model analyzes the disaster causing factors step by step;the neural network model is established to transform the qualitative and quantitative factors and their order eigenvalues,and the model is trained as the input of the neural network,so as to establish the nonlinear time series model between the disaster causing factors and the fire. In order to verify the effectiveness of the model,the simulation experiment is carried out based on real data. The results show that the neural network can effectively improve the accuracy of the correlation between various factors and fire warning after modeling the time sequence and nonlinearity of various factors.