为提高恐怖袭击应急管理的效率,设计了恐怖袭击的风险评估和预测系统.评估模型通过因子分析方法计算各类目标的相对风险指数,评估指标包含“威胁”、“脆弱性”、“后果”三大因素,具体数据从全球恐怖主义数据库(GTD)中进行采集.预测模型通过神经网络实现风险指数的预测,由于BP神经网络的梯度下降算法收敛较慢且易陷入局部最优点,因此利用遗传算法对神经网络的初始权值阈值进行优化,并提高预测精度.最后,对GTD数据库中的21类主要袭击目标进行算例分析,验证了该模型的可行性和准确性,同时还根据这些目标的风险指数进行原因分析和策略建议.%An evaluating and predicting system of terrorist attacks is designed to improve the efficiency of emergency management.In evaluating model,factor analysis (FA) is used to calculate the risk index among different attack subjects.The evaluating index contains 3 factors as threat,vulnerability and consequence,the related data are all collected from GTD database.In predicting model,neural network is applied to predict the risk index.However,as there are some flaws of BP neural network,such as slow convergence and easy of falling into local optimal,a genetic algorithm (GA) is applied to improve prediction accuracy through optimizing neural network's initial weights and thresholds.Finally,this model is applied to evaluate and predict the risk of 21 types of main targets recorded in GTD,the numerical example proves feasibility and accuracy of the model,also policy recommendations are provided at last.
展开▼
