사무실의 피난소요시간 산정을 위한 인공신경망 모델 개발

The simulation program to determine the required safety egress time(RSET) from an office room to evacuation zone has been devel-oped by using neural networks. The RSET for an office room in K office building at N city was computed based on the neural net-work and compared with that based on a traditional evacuation simulation program. The maximum error of the neural networks is5.89% about various occupant densities and 10.98% about various exit widths. The RSET with the conventional evacuation simu-lation program shows discontinuity curve, but the continuity curve which is generally presented in an engineering problem wasobtained with the program with the neural networks. The consistent results with the neural networks subjected to same conditions arecomputed so that the results are reliable and stable. The conventional evacuation program is time-intensive, but the developed neuralnetworks is capable of fast calculation within 1/100 sec. regarding a new condition. Therefore, The neural networks will enhance therescue activity and disaster management with real-time decisions for evacuation safety.