Simulation Of Flood Disaster Risk Based On Multi-Agent

Floods are main natural disasters, which become constraint for the development of our society, and cause serious casualties and loss of properties in the world. When flood happened, the government must adopt the rescue measures immediately according to the disaster trend, risk, and disaster affection. Flood disaster risk assessment is an important basis of disaster rescue for governments. The determination of disaster-affected populations and disaster situations are primary task for flood disaster risk assessment. Therefore, the disaster-affected population risk assessment and prediction of flood disaster are the foundations of natural disaster management and disaster rescue.It is highly difficult to analyze flood disaster by the field experiments because flood disasters are typical unconventional emergencies. Fortunately, the modeling and simulation are becoming effective means for understanding and transforming the world. From the view of system theory, flood risk assessment is typical complex system. With gradually thorough understanding of flood disaster risk, multi-agent simulation and modeling technology are introduced to forecast the risk of flood disaster, which becomes the trend for flood affected population risk analysis and assessment. The multi-agent modeling based flood disaster simulation can effectively integrate these three categories risk analysis models and methods for risk analysis and assessment of flood-affected populations.In order to meet the requirement of flood disaster risk assessment study, considering the characteristics of multi-agent modeling technique, we constructed a simulation system of flood disaster risk assessment based on multi-agent theory. The simulation framework of flood disaster risk assessment is also given. The architecture of single agent and multi-agent are designed, and communication mechanism among multiple agents is developed.Based on the simulation framework and models, we selected NetLogo platform to simulate the resident removal behaviors in different rainstorm flood disaster periods including earlier stage, interim stage, and later stage. Then, we analyzed the flood-affected population risk during rainstorm flood disaster process. The experiment results demonstrate our approach effective and practical.This research is significant for the theory and methodology of flood disaster risk assessment. It has a wide range of application prospect in natural disasters emergency relief, and will benefit both the society and the economy.