Simulation-based Emergency Evacuation Planing Using Population Dynamics And Mobile Guide

The emergent events in areas of high population density are likely to result in heavylosses, generating emergency evacuation plans effectively becomes one of the critical steps toreduce losses.. With the development of computer technologies, individual-basedmicrosimulation has become a major approach to simulating evacuation processes. However,the enlargement of evacuation population size will cause a dramatic increase in computationalload of microsimulation as well as the problem space of searching the optimal evacuationplans, which will then results in an increase in time of evacuation planning. Thetime-comsuming calculation of evacuation plans is not able to meet the timely responserequirement in urgent situations.. Therefore, it is necessary to study approaches to shorteningthe generation time of evacuation plans based on microsimulation.With the background of mobile guide, this study uses the patterns of dynamic populationdistributions to develop a knowledge database of evacuation plans, which will reducesolution-search space and thus shorten the computation time of generating near-optimalevacuation plans based on microsimulations. Particular.ly, this study develops a multi-agentbased evacuation simulation model based on dynamic population distributions. Using theevacuation empty time as the optimal objective, we choose the genetic algorithm to search outthe near-optimal evacuation plan. Then, we develop a knowledge database of evacuation plansby calculating the near-optimal solutions based on the typical pulation distributions of24hoursin a normal day. With a given population distribution in disaster events, we will search thetypical population distribution which matches the given population distribution best in theknowledge database. Then we use the evacuation plan based on the most similar typicalpopulation distribution as the initial seed of the genetic algorithm, which can speed up theevacuation planning process.This study takes the Huaqiangbei business district with the highest population density inShenzhen as the study area for experiments. The call detail records of this area are used togenerate the typical population distributions of24hours. We then establish a knowledgedatabase of evacuation plans by calculating the evacuation plans for typical populationdistributions of24hours respectively. Finally, we test the computation time of generatingevacuation plans with and without the help of knowledge database. The experiment resultsshow that the method proposed in this study can reduce the evacuation planning time by morethan80%with a guarantee of generating an acceptable near-optimal solution.