A Research Method for Early Warning and Response to Security Risks of Large Sporting Events Based on Artificial Intelligence Deep Learning Networks

Large-scale sports events are characterized by multiple factors, multiple levels and multiple targets, which imply large-scale security problems while generating exciting sports content. With reference to the results of two rounds of expert consultation, this paper proposes an early warning system for security risk of large-scale sports events structured by 9 secondary indicators and 27 tertiary indicators in three dimensions: pre-game preparation risk, game-time operation risk and post-game recovery risk. The principal component analysis is chosen as the calculation method of the principal component values of the experimental samples, and combined with the improved BP neural network algorithm, it forms the construction scheme of the risk warning model. The comprehensive weights of the early warning system indicators are calculated by integrating the Analytic Hierarchy Process (AHP) and the entropy weight method. Among them, the comprehensive weight values of the four secondary indicators, namely "venue security prevention risk", "venue operation plan risk", "venue financial preparation risk" and "venue post-event financial risk", are greater than 0.100, which are the key focus directions for security risk prevention in large-scale sports events. The constructed security risk warning model for large-scale sports events has an overall output accuracy of 91.07%, which can accurately output the security risk situation of sports events. This paper suggests that the preparation and organization of large-scale sports events should be supported by the early warning model, combined with the venues where the events are held, to formulate adequate security prevention and operation plans, prepare sufficient financial resources, and strategically respond to the security risks that may arise in sports events.