Predictive model of artificial neural network for disaster prevention

In order to predict the influence of earthquake on building and buried pipeline, predictive model is constructed on the basis of artificial neural network (ANN). According to double parallel feed-forward neutral network model, which is a basic model of Back Propagation (BP) network, predictive model and calculating method are analyzed. The model is applied to the calculation of earthquake affecting coefficient and the prediction of buried pipeline damage. Earthquake affecting coefficient is the ratio of maximum absolute acceleration of building and acceleration of gravity, which represents the earthquake influence on building. As an example application, the distribution of earthquake affecting coefficient in Tangshan City is calculated precisely. MATLAB is applied to analyze model structure, concealed layer number, neuron number of concealed layer, and training function. Finally, calculating results are analyzed and some advice is proposed for artificial neural network modeling.