Modeling Evacuation Behavior And Analyses Of Dynamic Behavior Characteristics Of Crowd

In large public places and large activities, pedestrians are always complex and in a high density. If emergency occurred, a small turbulence may lead to an unstable state, or even serious stampede. Pedestrian evacuations have been widely concerned by scholars. It is very useful for managers to study the dynamic movement behavior of pedestrians. Making scientific evacuation strategies and effective controlling policies are the keys to reduce the accidents and guarantee the safety of pedestrians. On the basis of the studies of evacuation experiments in traditional scenarios, this paper designs pedestrian experiments in complex scenarios and makes qualitative and quantitative research of the characteristics of microscopic movement behavior and macroscopic route choice behavior of pedestrians. This paper studies the approaches of modeling pedestrian evacuation process and dynamic behavior characteristics based on the heuristic force-based model and user optimal criterion. The main theses of this research are as follows:(1) According to the current experimental research of pedestrian evacuation dynamics, a new series of experiments are conducted by considering the route choice behavior. The scenario composes of multiple exits, routes and moveable obstacles. In our experiments, the obstacles are rearranged to resize the sizes of all routes, and to make analyses of the impact factors of route choice behavior. Personal information and individual route choice results are acquired according to questionnaires. By extracting individual motion trail from the experimental videos, the characteristics of evacuation dynamics are obtained, such as the numbers of pedestrians choosing each route, the evacuation time of each exit, and flow distribution in scenario of each time. The experimental results show that pedestrians take into account some factors when making route choices, such as the distance between current location and destination, the width and length of each route and exit, the flow distribution of each route. Pedestrian route behavior has regional, stochastic and dynamic features. The route choice behavior basically satisfies the user optimal criterion.(2) Based on heuristic force-based model, the mathematic features of heuristic function and the mechanism of overtaking front velocity-obstacles are analyzed. A heuristic detouring algorithm (HDA) is proposed to efficiently solve the non-linear and non-monotone heuristic function. In general, an approximate solution will be obtained by HDA; in the situation of one standing obstacle in vision field, the solution will be accurate. Applying four methods that are social force model (SFM), optimal reciprocal collision avoidance (ORCA), enumerative algorithm (EA) and HDA to simulate pedestrian flows, the accuracy and efficiency of the proposed HDA are tested and validated. The simulation results show that the HDA has nearly the same accuracy with EA, but the efficiency is about20%higher than EA. Compared with ORCA and SFM, the fundamental diagram obtained by HDA consists with reality.(3) A modified heuristic force-based model is proposed to describe staircase movement. According to empirical and experimental data, the proposed model is validated by the features of staircase movement. Considering the geometrical and computational factors, the three-circle shape is used to describe the body shape of each pedestrian, and the calculation method of optimal desired velocity is also proposed. The model takes into account the effect of physical restriction, gravity and geometrical size. Introducing a linear variable relaxation time, the model can describe the staircase movement in different crowd densities and movement directions. Compared with measurement data in Beijing Railway Line1, the simulation results well reproduce the actual data. Microscopically, the model can obtain individual walking speed and inflow-outflow rate function, and the relative error is about7.26%. Macroscopically, the model can reproduce the line forming phenomena, obtain the evacuation time, and estimate the capacity of staircase.(4) Considering the microscopic and macroscopic features of evacuation behaviors, a micro-and macro combined model is established to describe the pedestrian evacuation behavior in buildings with inner structures. Microscopically, the heuristic force based model is introduced to describe individual movement behavior; macroscopically, the equilibrium theory used in traffic network design is included to model the route choice behavior. The result of microscopic model is individual speed and location, and then integrated into average flow and speed in each path, which will be used for the pedestrian to make route choice. The result of macroscopic model is discretized to individual direction of desired velocity. A series of simulations have been carried out, and the numerical results have been compared with experimental data and empirical data. The results show that the proposed model can qualitatively and quantitatively reproduce crowd’s dynamic evacuation process. The route choice results obtained by simulations are basically consistent with actual data, and the average relative error of the route choice results is about8%. In addition, the model is applied to make a large scale simulation in LOTTEMART supermarket in Sidaokou, Beijing. The effect of emergency to evacuation process is analyzed, and a method of identifying congested regions is proposed.(5) Considering the differences of individual behavior, a multi-class dynamic user optimal model is established. Meanwhile, a microscopic model including the herding effect and information spreading process is proposed. According to the analysis of evacuation network characteristics, a general framework of evacuation network is established. Macroscopically, pedestrians are divided into different classes by their obtained information about exits. By simulations, the effect of pedestrian numbers on evacuation efficiency is analyzed. The results show that the pedestrians with perfect information can improve the evacuation efficiency. In a crowd, a small part of these pedestrians can lead to an approximately optimal status. Microscopically, three traditional herding behaviors are included in the route choice model. The herding behaviors are composed of’sub-group’ behavior, herding behavior without information spreading, and herding behavior with information spreading. According to simulations, the results show that in a low density, pedestrians can find a route quickly by herding behavior; however, in high density, the herding behavior may lead to a local congestion, and reduce the evacuation efficiency.