Abstract:
An evaluation of Required Safe Egress Time (RSET) is an important step in performance based on fire safety engineering. Pre-evacuation time is currently regarded as the key critical evacuation process. If pre-evacuation time is not explicit, RSET timeline will not be reliable and effect on life of evacuees in the building. Therefore, this study attempted developing the model for predicting pre-evacuation time in plastic industry. Applying the Stepwise Multiple Regression (MRA) to find an affected input variables and constructed forecasting two pre-evacuation time model were Stepwise Multiple Regression (MRA) and Back Propagation Artificial Neural Networks (BP-ANNs : 4-10-1). Finally, the comparison between the two model, the error forecasting model of MRA and BP-ANNs (4-10-1) showed the Mean Absolute Percentage Error (MAPE) of 4.43 and 1.64 respectively. Hence, BP-ANNs 4-10-1) is an optimal model for predicting pre-evacuation time. This finding shows the advantage of BP-ANNs (4-10-1) is more suitable (comparing with MRA) to predict fire evacuation plan and eliminate factors that could delay evacuation time in plastic industry.