城市居民小区地震火灾发生率的非线性预测

利用人工神经网络对地震火灾发生率进行非线性预测具有实用价值和理论意义。分析了人工神经网络预测的优势和不足，并提出算法改进意见，在综合分析人工神经网络预测的工作机理、构造方式以及影响城市居民小区地震火灾发生率的主要因素的基础上，构造人工神经网络模型进行火灾发生率的实例预测，并针对预测结果分析了各影响要素对地震火灾发生率发生影响的强弱程度，非线性预测的结果与预期结果相符。

The Nonlinear Prediction to the Seismic Fire in the Urban Area

To predict the seismic fire in an urban area,this paper tries to put forward a nonlinear means-Artificial Neural Network(ANN).It analyzes the advantages and disadvantages of ANN for predicting,and proposes a set of mathematical formula and some suggestions for further improvement.Based on these analysis and the structure of ANN and some factors for the seismic fire in urban areas,a software-Nshell is used to predict the fire frequency.According to the results available,the paper also studies how the factors concerned influence on seismic fire in urban areas.Experiments show that the results obtained through the non-linear means correspond to the predicted results.