A Bayesian network model for the Asian games fire risk assessment

Song Lu,Dan Wu,Lu Shi-chang,Zhang He-ping

DOI: https://doi.org/10.1109/ISCRAM.2011.6184131

2011-01-01

Abstract:In order to assess the fire risk of the gymnasiums, the assigned hospitals and hotels for 2010 Guangzhou Asian Games, a Bayesian network (BN) fire risk assessment model was developed. The BN model consists of 45 nodes that are divided into four subnetworks: fire safety management, fire-fighting facilities, evacuation and fire hazard assessment. The fire safety management subnetwork specifies the influence of fire safety management on fire risks; the fire-fighting facilities subnetwork predicts the working conditions of fire-fighting facilities; the evacuation subnetwork simulates the required safety egress time; and the fire hazard assessment subnetwork predicts fire hazard results based on the results of above three subnetworks. Important risk factors due to the features of the Asian Games include security funding, police reinforcement, secondary and derivative events. The model was applied to 50 gymnasiums, 57 hotels, 35 hospitals and 2 congregations (opening and closing ceremony) before the Asian Games, and the assessment results were used in reducing and controlling fire risks. The practical working procedure of the Asian Games fire risk assessment work is also presented. The application of this model shows that it not only can perform fire risk assessment but also can conduct fast disaster situation assessment based on observational data during a fire. © 2011 IEEE.