Rapid Risk Assessment of Emergency Evacuation Based on Deep Learning

To address the continuous occurrence of safety accidents in large public buildings, emergency evacuation has been an essential means of emergency disposal. However, risks also exist in the evacuation processes. Evaluating the risk of evacuation processes can be used for improving the safety of the evacuation processes and providing additional support for evacuation decision-making, which has important practical significance. At present, because of the complexity of evacuation processes and the lack of data, the research on evacuation risk assessment is still limited. Traditional risk assessment methods have more subjective and are difficult to fulfill the requirements of timeliness in emergency evacuations. With the development of artificial intelligence, it has provided a possibility to use deep learning methods to excavate the internal relationship of complex evacuation systems and achieve rapid risk assessments. This article innovatively applies deep learning methods to the field of risk assessment of evacuation. An approach based on the convolutional neural network is proposed in this article to establish an evacuation assessment model. Two network structures, Lenet and Resnet, are selected to train the model, respectively. A real case of the large stadium is used to illustrate the assessment way, and a large number of experiments were carried out to obtain the data required for training. The result shows that the deep learning method can realize an efficient and fast risk assessment.