随着我国大、中城市中高层建筑越来越多、越来越高，其安全问题也愈加突出。一旦高层建筑起火，火与烟蔓延速度快，扑救困难大。为了减少人员伤亡数量，需要对建筑内人群进行疏散。建筑中传统的疏散方式是楼梯，但大量的疏散者使用楼梯，会导致楼梯拥挤，延误疏散时间。如果疏散者可以使用电梯，并充分利用楼梯、电梯和避难层（即多模式），使多模式疏散相互协调、配合，则可以有效地提高疏散效率。因此，本文以高层建筑中人群的多模式协同疏散为切入点，开展高层建筑疏散实验，研究楼梯和电梯的人员疏散行为和人群疏散规律，基于从已有研究和实验中得到的人群疏散行为构建多模式疏散仿真模型，并根据仿真对建筑多模式协同疏散方案进行优化。基于已有的对人群疏散行为的研究，进一步开展疏散演习和实验。对两个高层建筑中人群疏散演习数据进行了采集。在多个因素对人群疏散行为有影响的假设下对疏散实验进行设计，而后开展三组实验。通过对实验数据的分析，（1）得到人群疏散时间、使用电梯与楼梯疏散的人数比例、人员等待电梯时间、电梯开关门时间和人员移动速度等数据；（2）发现烟气对人员使用电梯疏散的行为以及对楼梯与电梯的选择决策影响较小，而等待电梯人数、疏散经验和人员社会关系则对二者影响较大；（3）验证了多模式疏散的可行性，并揭示了多模式疏散存在的问题。实验结果对建筑安全设计，尤其是电梯的使用有着一定的参考价值。耦合在已有研究和实验中得到的人群疏散行为，构建高层建筑人群多模式疏散仿真模型：（1）考虑人员使用电梯疏散的行为和火灾危险度，对电梯调度进行优化，提高了电梯疏散效率；（2）考虑楼梯结构、人员行走偏好、人员生理条件和人员心理状态等因素，改进了基于元胞自动机的楼梯仿真；（3）通过改变楼梯疏散仿真空间结构以及简化人员移动规则，降低了仿真计算复杂度；（4）利用疏散演习数据从多角度验证了仿真的有效性。针对高层建筑人群疏散的特点，基于多模式协同疏散仿真，（1）建立高层建筑多模式协同疏散控制系统；（2）提出分步式实时优化方法，对高层建筑疏散方案进行优化；（3）对多模式协同疏散方案优化问题进行建模并求解。最后，在假设人群服从疏散指示的情况下，根据多模式协同疏散方案对人群进行疏散指示，提高人群疏散效率。
As the number and the height of high-rise buildings increase, the building safety problem is becoming more crutial. Once a high-rise is on fire, the fire and smoke propogate very fast, and it is difficult to extinguish it. In order to reduce the number of casualties, people should evacuate out of the building as soon as possible. The main exit is the stair, but if everyone use it, conjestion may happen and the evacuation time may be delayed. If evacuees can use safety elevators, and make full use of stairs, safety elevators and refuge floors (multi-mode), the efficiency of evacuation will be improved. In this paper, take multi-mode collaborative evacuation in high-rise buildings as the key point, we carry out evacuation experiments in high-rise buildings to study crowd movement patterns of using stairs or elevators. Take the advantages of evacuees’ behaviors from both existing studies and our experiments, we establish a multi-mode collaborative evacuation simulation model. Then we optimize the multi-mode collaborative evacuation plan based on evacuation simulation.Based on the advantages of existing studies, we develop evacuation drills and experiments to further study crowd behaviors. We collect the data of two drills in high-rise buildings, and develop three groups of experiments considering several factors. According to the analysis of experimental data, (1) we get the crowd evacuation time, the ratio of using elevators and stairs, the waiting time for the elevator, the time to open and close the door of the elevator and evacuees’ speed; (2) we found that smoke have a little impact on evcuees’ behaviors and the choice between stairs and elevators, and the number of waiting evacuees, evacuation experience and social relation will influence them a lot; (3) we verify the feasibility of multi-mode evacuation, and reveal the problems of multi-mode evacuation. The experimental results are helpful to building safety design, especially to the design of safety elevators.Considering evacuees’ behaivors from both the existing studies and our experiments, we establish a multi-mode collaborative evacuation simulation model: (1) considering the behaviors of using safety elevators and the risk of fire, we optimize the elevator scheduling problom, and improve the efficiency of the elevator evacuation; (2) considering the structure of a staircase, evacuees’ walk preference, physical and psychological status, we improve the stair evacuation simulation based on cellular automata model; (3) according to the re-design of simulation space, we simplify the evacuees’ movement rules, and reduce the simulation computation complexity; (4) we use several evacuation drills to validate the effectiveness of the simulation from multiple perspectives.Based on the multi-mode collaborative evacuation simulation and evacuees’ behaviors during high-rise building evacuation, (1) we establish a high-rise building multi-mode collaborative evacuation control system; (2) we introduce a step real-time –by-step optimization method, and optimize the evacuation plan based on such method; (3) we establish the model of the optimization of multi-mode collaborative evacuation plan and solve it. Finally, if we assume that evacuees will follow the guidance, the efficiency of evacuation will be improved based on the optimized multi-mode collaborative evacuation plan.
