基于小波神经网络的多楼层疏散模型

Multi-floor Evacuation Model Based on Wavelet Neural Network

多楼层环境下室内人群疏散问题是社会关注的热点,而传统的社会力模型在模拟多楼层环境时容易出现停滞等待现象。基于小波神经网络来改进社会力模型,建立一种新的多楼层疏散模型。该模型利用场域模型来获得行人的运动方向,以此作为社会力模型中行人的自驱力方向。同时给出了多楼层环境下出口拥挤度、路径拥挤度和平均速度的评价指标,并利用小波神经网络建立疏散优化方法。利用搭建的仿真平台和上述改进模型模拟了多楼层疏散过程,深入分析了影响该模型的关键因素。该环境下疏散结果表明:适当提高行人的疏散速度有利于提高疏散效率,但是速度过大会使行人快速聚集在楼道处,反而不利于疏散;此外疏散时间随楼梯宽度的增加呈现递减趋势直至平稳,当楼梯宽度达到8 m时,再增加楼梯宽度也不能降低疏散时间。

Crowd evacuation in a multi-floor environment is a popular social concern, while the stagnation phenomenon easily occurs when simulating a multi-floor complex environment with the traditional social force model. Therefore, An improved social force model is proposed by a wavelet neural network, and a new multi-floor evacuation model is built. In the model, a pedestrian’s direction of movement is obtained by the field model, which is used as the self-driving direction of the social force model. Meanwhile, the evaluation indexes of the exit congestion degree, path congestion degree, and average velocity in a multi-floor environment are given, and a wavelet neural network is employed to develop an evacuation optimization method. The evacuation process is simulated by the platform and the improved model, and the key factors in this model are studied. The results show that properly increasing the evacuation velocity of pedestrians can improve evacuation efficiency, but if the velocity is too high,pedestrians will gather in the corridor quickly, which is not conducive to evacuation. In addition, the evacuation time shows a decreasing trend with the increase in the staircase width before becoming stable,and when the staircase width reaches 8 m, further growth of the staircase width will not reduce the evacuation time.