Prediction of Terrorist Attacks in China based on BP improved Algorithm and GTD

Abstract

Abstract. Due to the different national conditions, the driving forces and factors of national terrorist attacks vary. Therefore, this paper takes GTD China sample data as the research object to study and predict. The prediction process is as follows: On the basis of the BP network-based model for predicting the most dangerous areas, combined with the GTD sample data, the best number of nodes in the implicit layer of the prediction model is automatically selected by combining the empirical formula with the MATLAB program. Three improved BP algorithms are used to train the network model. The results show that the training error of Levenburg Marquardt algorithm is minimal and the convergence speed is fastest. Through the training and simulation of the model, it is proved that the model has high precision and can meet the requirement of practical application.