拥挤系统阻碍物位置的遗传算法优化

Using genetic algorithm to optimize obstacle location in a crowded system

首先对Helbing等人建立的社会力模型进行了适当修正,建立了人群运动仿真模型;然后采用实数编码遗传算法并设计了自适应的交叉及变异策略,建立了优化阻碍物位置的数学模型;最后通过高性能计算平台进行仿真计算,得到了最有利于提高拥挤人群疏散速度的阻碍物设置.通过优化房间和人群所形成的拥挤系统中阻碍物的大小、数量和位置,有效提高了恐慌发生时房间内人群的逃生效率,该结果对建筑设计具有重要参考价值.

The basic social force model was modified to build a pedestrian dynamic model.Real-coded genetic algorithm with self-adaptive cross-over and mutation strategy was used to optimize obstacle setting.High performance computation platform was used to simulate and find the best setting of obstacles.Obstacle size,number and position in the crowd system containing pedestrians and rooms were optimized to improve pedestrian escape speed in a panic situation.These results have practical meaning for architectural design.