Research And Implementation Of Intelligent Screening Method For Urban Anti-Terrorism Emergency Factors Based On Ensemble Learning

The response and handling of terrorist incidents has always been a necessary measure to maintain national security.Local governments often need to carry out normalized training and rehearsals,and make relevant counter-terrorism emergency response and resource allocation plans based on different situations of imagined terrorist attacks to deal with possible threats.However,there may be dozens of factors that affect an event,but only a small number will have a significant impact on the outcome.Therefore,it is necessary to screen out the emergency factors that have an important impact on urban anti-terrorism emergency operations.This will not only help government departments to allocate resources and improve efficiency during normal training and rehearsals,but also make use of urban anti-terrorism emergency responses when terrorist attacks occur.The simulation system quickly makes counter-terrorism emergency decisions.The thesis proposes an intelligent screening method for urban antiterrorism emergency factors based on ensemble learning.The method first uses TopSis method and entropy weight method to extract a variety of indicators in the result of anti-terrorism emergency actions into a comprehensive indicator,and then improves the shadow factor construction method in Boruta algorithm for this scenario,so that it can deal with the interaction between emergency factors in urban anti-terrorism emergency actions.The random forest and the improved Boruta algorithm are used to screen the emergency factors,and the emergency factors that have an important impact on the action results are obtained.Finally,based on the method proposed in the thesis,a prototype system for intelligent screening of emergency factors is designed and implemented.Using the simulation data provided by cooperative enterprises,the thesis compares the proposed intelligent screening method for emergency factors with the traditional factor screening algorithm in terms of the number of experiments,monotonicity requirements and the number of processing factors,and shows obvious advantages.Then,comparing this method with the random forest algorithm without Boruta algorithm,and AdaBoost algorithm,the accuracy rate is significantly improved,which indicates the effectiveness of this method in the experiment.