Research On Safety Risk Assessment Of Large Venues Facing Vehicle Collision

With the accelerating progress of China’s going global,China is hosting an increasing number of major activities while vehicle collision has become one attack means favored by terrorists as there are strict regulations on the management and control of such violent tools as guns and explosives in China.Large stadiums,characterized by excessive gathering of crowds and extensive attention paid by media at home and abroad,are very likely to become the object of attack from vehicle collision.Once the events of vehicle collision occur in the large stadiums,it’s hard to estimate the loss of personnel,property and reputation.Therefore,it’s of great significance of research to conduct a targeted risk evaluation of this kind of threat faced by large stadiums,which can help prevent the events of vicious vehicle collision.Taking the large stadiums as the research object,this passage studies the risk evaluation of vehicle collision and introduces the module of regional abnormal behavior analysis to build the risk evaluation indicator and the evaluation model from possibility,regional abnormal behavior,vulnerability and aftermath.Combined with the characteristics of the attack from vehicle collision,the research creatively introduces the mechanical learning algorithm into traditional risk evaluation methods and build a risk evaluation model integrating subjectivity with objectivity,achieving the effect of qualitative and quantitative evaluation.Then,it designs and develops a reproduceable and promotable risk evaluation software on the basis of the model of various modules.In the end,it selects the real stadiums to apply and test them by taking some real major activities as the scenes,and puts forward some targeted suggestions to improve the preventive and protective system.The main work of this passage is given as follows:Collected and sorted out the relevant data of large venues suffered from vehicle collision attacks at home and abroad,used the Text Rank algorithm to extract the keywords of vehicle collision events,established safety risk assessment indicators for large venues oriented to vehicle collisions,including 4 first-level indicators and 47 final-level indicators,and determined the quantitative and grading standards of the final-level indicators according to relevant standards.In the stage of risk analysis,this paper constructed A dataset for vehicle collision likelihood analysis,and used the ADASYN algorithm for data balance and data expansion;constructed possibility analysis models independently by four machine learning algorithms of KNN algorithm,CART decision tree algorithm,BP neural network,and particle swarm optimization BP neural network,namely PSO-BP neural network,and analyzed and compared the training results of the four models;constructed the regional abnormal behavior analysis model,by the integral early warning method through the regional abnormal behavior data;respectively constructed the vulnerability analysis and consequence analysis models by the G1+CRITIC combined weighting method;finally,a risk assessment matrix is constructed to complete the risk assessment and processing.After completing the risk assessment model,this paper designed and implemented a largescale venue safety risk assessment software for vehicle collision,which integrated the possibility analysis model,regional abnormal behavior analysis model,vulnerability analysis model and consequence analysis model into the software system.In order to verify the feasibility and rationality of the risk assessment process and risk assessment software,this paper selected an Olympic sports center as the assessment target,and took an international sports event as the scene to verify the application of the above risk assessment process and method.