To understand evacuation behavior and psychological characteristics in the event of a sudden fire in an ancient town，and to accurately identify the factors influencing personnel evacuation behavior，a questionnaire survey on evacuation behavior was conducted in an ancient town in Zhejiang Province. The survey focused on three aspects：basic population characteristics，awareness and attitudes toward fire risk in the ancient town，and evacuation behavior. Subsequently，the correlation between evacuation behavior and factors such as age group，frequency of fire drills，and understanding of fire risks in ancient towns was comprehensively analyzed using the Pearson's Chi-square test and eXtreme Gradient Boosting (XGBoost) machine learning algorithm. The results indicate that tourists (77.78%) are more aware of fire risks in the ancient town than local residents and merchants (22.22%). Age has a significant impact on returning behavior. The higher the level of education，the less likely respondents are to engage in overtaking behavior during evacuation (decreasing from 85% to 33%). Individuals who have not proactively familiarized themselves with emergency exits are more likely to exhibit herd and 'habitual' behavior when choosing evacuation exits (increasing from 8% to 33% and from12% to 20%，respectively). Special attention should be given to fire safety training for local residents and merchants. Fire safety management in the ancient town should be strengthened，and evacuation plans tailored to different age groups should be developed to improve evacuation capabilities in the event of a sudden fire in the ancient town.

Based on the high and steep slope project of an open-pit slope in cold region，30 freeze-thaw cycle tests were conducted. The temperature range was set from -30 ℃ to 20 ℃.Subsequently uniaxial variable upper limit cyclic loading-unloading tests as well as synchronous acoustic emission monitoring tests were carried out. Slope rock masses with fracture angles of 0，25，50 and 75° were used in potential slip zone. The freezing-dynamic (freeze-thaw cycles and cyclic loading and unloading) combined damage and deterioration characteristics and mechanical properties of slope rock mass were explored in macro and mesoscopic scales. Furthermore，the crack initiation，propagation and failure modes of fractured rock mass were studied. The results show that as the fracture angle increases，the freeze-thaw damage effect on the fractured rock mass gradually decreases，while the compressive strength and elastic modulus exhibit a linear increasing trend with the maximum deformation of fatigue resistance of 0.558 3% at 75°. Compared to ordinary uniaxial loading，the compressive strength of fractured rock masses under cyclic loading and unloading condition decreases by 5.6 MPa. The Felicity ratios of different rock masses decrease with the increase of cyclic levels，and the Felicity ratios at the final failure stage were all below 0.7. As the cyclic loading level increases，the increment of cumulative dissipated energy decreases with the increase of fracture angle. The rock masses mainly exhibit tensile failure，but when the angles exceeded 25°，there is a trend of transformation from tensile and mixed failure to shear failure.

In order to effectively improve the comprehensive prevention and control ability of coal mine water disaster accidents，the complex network and robustness analysis method were used to explore the key causes of accidents. According to the report of 111 typical coal mine water disaster cases in China in the past 40 years，an unweighted directed network of coal mine water disaster was constructed. The weights of each cause，such as degree，clustering coefficient and betweenness centrality at different scales of the network were calculated，and the key causes were excavated. The damage of intentional attack to the cause network under three different strategies was evaluated，and the optimal solution of systematic prevention and control was sought. The results show that the coal mine water disaster cause network has the characteristics of small world，and the causes are closely related. The attack ranked by degree has the greatest damage to the model. The corresponding key causes include organizing workers to carry out risky operations，failing to implement the management responsibility of water prevention and control technology，failing to carry out water exploration and drainage work as required，illegal organization of production operations，imperfect safety management institutions，inadequate hidden danger investigation，loopholes in coal mine safety supervision and supervision，and failure of higher-level units to perform safety management responsibilities.

To investigate the methane DDT distance and maximum explosion pressure (P_max) in shale fractures，a multi-scale adjustable 3D planar slit detonation system was developed. Experiments with methane-oxygen premixed gas under 4 different hydraulic diameters，along with numerical simulations，were conducted to examine shale gas combustion under high pressure. Results show that methane-oxygen premixed gas can sustain self-propagating explosion within a hydraulic diameter range of 1.9 to 11.43 mm. Both P_max and peak pressure rise rate increase linearly with initial pressure. Under a hydraulic diameter of 11.43 mm，P_max closely approaches theoretical detonation pressure. As the hydraulic diameter decreases，the P_max-to-initial pressure ratio decreases. The initial pressure and the DDT distance follow a power-law relationship. Increasing the initial pressure or reducing the hydraulic diameter can shorten the DDT distance，thereby accelerating the DDT. The simulation shows that methane-oxygen premixed gas explosions can produce an overpressure of 330 MPa，capable of fully fracturing rock cracks.

In order to reduce the risks associated with the continuous growth of airport flight area size and flight volume，the safety resilience assessment of airport flight areas was carried out. First，risk factors were identified by analyzing the historical data of airport flight zones. Second，key risk factor weights were quantified，and a SD-based safety resilience assessment model for airport flight zones was constructed to propose safety resilience indicators. Then，the safety resilience of airport flight zones was assessed through simulation analysis，and targeted enhancement strategies were proposed. Finally，a large domestic airport flight area was taken as the research object to assess its safety resilience. The results show that among the personnel factors，the performance of the flight crew has the greatest impact on the level of operational safety resilience. By controlling the flow in the controlled airspace，enhancing safety awareness and increasing management inputs，the operational safety resilience of the flight area is improved by 9.11%. Among the environmental，equipment and management factors，the degree of improvement of the equipment updating mechanism has the greatest impact on the operational safety resilience level. By accelerating the frequency of equipment renewal，improving equipment deficiencies and increasing management inputs，the operational safety resilience of the flight area is increased by 21.49%.

To address the issues of complex environmental interference and low recognition rates in early-stage tunnel fire detection，an improved YOLOX-based detection method，YOLOX-T，was proposed. The proposed method incorporated a NAM into the YOLOX network to suppress environmental noise and interference，thereby enhancing the model's robustness. A weighted BiFPN was integrated to improve multi-scale feature extraction and fusion. Furthermore，an α-IoU(Intersection over Union) loss function was employed to enhance the detection accuracy of early-stage tunnel smoke and flames，which often exhibit indistinct contours. Addressing the scarcity of publicly available datasets，a tunnel fire dataset encompassing both real-world and simulated scenarios was constructed through web data acquisition，simulated fire experiments，and the augmentation of existing datasets. Experimental results on the self-built dataset demonstrate that，compared to the original YOLOX model，the YOLOX-T method achieves improvements of 1.89% in mean Average Precision (mAP@0.5)，0.88% in mAP@0.5~0.95，4.57% in precision，and 5.45% in recall. The improved algorithm can achieve better detection performance.

In order to reduce the various risks associated with storm and flood hazards，an ontology model of storm and flood hazard risk is proposed. Storm and flood disaster risk elements were identified from five aspects：environmental factors，equipment and facility factors，management factors，human factors，and information factors. The concepts of storm flood class and state space were given respectively，and data attributes and object attributes of each hazard element are defined. And visualization of the storm flood ontology model based on Protégé application. The results show that the storm and flood ontology model can express the risk elements and attributes clearly and accurately，and retrieve the links between the risk elements scientifically and quickly，so as to help the decision-makers respond effectively. Providing a data storage and retrieval platform for heavy rainfall and flood disaster risk assessment，it can realize user knowledge reuse and sharing，and provide a reference for emergency response and scientific decision-making by the government and emergency management authorities at the same time.

To grasp the development trends in urban emergency management research under extreme circumstances，a systematic review of the main studies and research characteristics in this field has been conducted. Using the scientific knowledge mapping tool VOSviewer software and Web of Science Core Collection (Wos CC) as the data retrieval source，this article performed a visual analysis on research literature on emergency management in extreme situations. The results show that in recent years，the research on urban emergency management in extreme situations has been increasing year by year，but the overall number of articles published is not much，with a total of 145 articles from 2004 to 2024. The United States，China，Australia are among those with the highest number of research papers，making significant contributions to the development of this field. Obvious research teams have been formed，mainly in China and the United States，but there is a lack of international cooperation. Through cluster analysis，the research topics can be summarized into five aspects，namely，disaster prevention management，disaster prediction models，urban vulnerability analysis，smart cities，and the impact of climate characteristics. Currently，as the probability of cities facing extreme risks increases，it is urgent to strengthen international cooperation，fully utilize the latest technologies such as artificial intelligence and big data，and construct a theoretical system for urban emergency management capable of facing extreme situations.

To evaluate the seismic resilience of urban group buildings under different demands，an integrated assessment methodology was developed for urban building clusters' earthquake resistance. A tree-augmented naive Bayesian method was employed to construct a BN model that rapidly predicted structural damage states under various target earthquakes through multi-parameter analysis including building types，construction eras，and structural configurations. Multiple resilience evaluation theories and methods-comprising the resilience index method，resilience element rating method，risk-resilience analysis method，and group buildings resilience indicator evaluation method-were integrated to comprehensively assess the seismic resilience of group buildings in a specific street of Fengtai District，Beijing. Comparative analyses were conducted to examine the advantages，limitations，and applicability of these approaches. The results indicate that the resilience index method proves suitable for rapid assessment of building recovery capacity，while the resilience element rating method facilitates intuitive visualization of structural resilience levels. The risk-resilience analysis method demonstrates superior capability in handling disaster uncertainties and stochastic characteristics，whereas the group buildings resilience indicator evaluation method emphasizes the critical influence of functional requirements on comprehensive assessments.

To gain a deep understanding and evaluate the command capability and current status of counties in response to flood disasters，this paper identified 12 major influencing factors through literature analysis. Social network analysis was conducted using Ucinet software，and Netdraw software was utilized for visualization to depict the social network relationship diagram among these influencing factors. By calculating the in-degree and out-degree，as well as centrality of the influencing factors，indicator classification was performed. Drawing on the maturity model of command and control capability，an assessment model based on social networks for the maturity of county-level flood disaster emergency command capability was developed. Puyang county was selected as the empirical research object. Through questionnaires and interviews，a total of 342 data entries from 24 townships in Puyang County were collected to evaluate the county's flood disaster emergency command capability. The results indicate that it is feasible to classify indicators hierarchically through social network analysis. The established indicator system is scientifically and accurately reflected in the comprehensive operational level of county-level flood disaster emergency command. The assessment model can effectively evaluate the maturity levels of emergency command capability.