In order to improve the evacuation efficiency of public buildings and reduce casualties caused by emergencies，in view of the optimization problems of emergency exits and evacuation plans，a mathematical model and algorithm for the single-source multi-sink evacuation problem were proposed. Firstly，all evacuation paths were identified using path algorithm and sorted based on evacuation time. Then，the recursive formula for calculating evacuation time was derived based on the User Equilibrium (UE) principle and k-shortest path. The capacity allocation issues of each evacuation path at intersections were analyzed. After that then the optimization strategy for emergency exits is obtained. Finally，taking a building as an example，the evacuation network was established and calculated to validate the effectiveness of the algorithm and optimization strategy. The results show that the evacuation efficiency of the building network increases gradually with the increase in the number of evacuees，and the rate of change stabilizes after reaching a certain threshold value，Evacuation efficiency is closely associated with the total traffic capacity of specific emergency exits. Through the optimization of specific emergency exits，the evacuation efficiency of the network is enhanced，evacuation time can be reduced，and the optimization effect becomes more pronounced as the number of evacuees increases.

To enhance the evacuation efficiency of deeply buried subway stations，a standard subway station was selected to establish an elevator-assisted evacuation model for deeply buried subway stations. The average evacuation time of passengers was selected as the primary evaluation metric. Variation characteristics in evacuation efficiency were calculated and analyzed under the combined influence of various factors，including buried depth of the subway，passenger flow intensity，the proportion of passengers opting for elevator evacuation，elevator operating parameters，the number of elevators，and acceptable queue size through simulation. The results indicate that the advantages of elevator-assisted evacuation are more pronounced when the subway depth exceeds 30 m. When passengers maintain their original evacuation paths，the evacuation time is inversely related to the proportion of passengers choosing to use the elevator during off-peak periods，but positively related during peak hours. Furthermore，when passengers alter their evacuation paths due to queue size，evacuation efficiency improves under different buried depth scenarios. In a subway with a burial depth of 90 m and an acceptable queue size of 30，the overall evacuation efficiency reaches its peak. When planning subway exits，reasonably increasing the number of elevators and their rated load，as well as operating speed，can effectively balance evacuation efficiency with cost control.

To investigate the research dynamics，hotspots，and frontier trends in the field of emergency supplies scheduling，the data sources of China National Knowledge Infrastructure (CNKI) and Web of Science(WoS) were used to search and filter 321 Chinese and 497 English articles. The bibliometric and knowledge mapping software were used to conduct basic feature analysis and development trend analysis. The results show that the number of articles in the field of emergency supplies scheduling both domestically and internationally has a wavy growth. The overall research is in the rapid development stage. In domestic core author teams，there is relatively low collaboration density. While in international contexts，cross-national and cross-regional academic exchanges are frequent，with China，Canada，and Singapore serving as the core. The research focus of both domestic and international studies is basically the same，mainly revolving around model design，optimization algorithms，path location issues，etc. However，international research has been earlier and more in-depth in studying the psychological perception of disaster victims.

In order to enhance the reliability and safety of gas network operations and improve the fault diagnosis capabilities for gas network leaks，while addressing issues such as the scarcity of real gas network leak data samples and variations in operating conditions，a gas network leak localization method based on transfer learning was proposed. Firstly，the Random Forest feature importance ranking method was used to select five pressure monitoring points in the TGNET simulation network. Subsequently，pressure monitoring point data under three different pressure conditions were respectively used as the source domain and target domain input features. The traditional JDA method of transfer learning was improved to reduce the feature distance between the source domain and the target domain. Furthermore，the CS algorithm was employed to optimize the dimensionality after mapping d' and the learning rate λ parameters of the improved transfer learning algorithm，ultimately achieving the diagnosis of unlabeled target domain leak segments. The results indicated that the proposed leak localization method for complex gas networks can effectively improve the localization accuracy of unlabeled gas network leaks，achieving higher accuracy compared to traditional.

To enhance the ability to cope with reservoir landslide hazard risks under extreme climate，a framework for multi-dimensional scientific observation and hydrometeorological early warning was constructed using multi-source monitoring data and machine learning algorithms. The spatiotemporal pattern and main controlling factors of landslide deformation were identified by analyzing the multi-annual observations of the two landslide cases，involving Sentinel-1，global navigation satellite system (GNSS) surface displacement and fiber optic (FO) strain. Leveraging the boosting decision tree (BDT) algorithm，a hydrometeorological early warning method based on slip zone real-time strain (RTS) was proposed，and the generalized framework of monitoring，early warning and emergency management strategies for reservoir landslides was systematically discussed. The results indicate that landslides with different deformation mechanisms show different spatiotemporal deformation characteristics，and landslide activities are closely related to localized anti-sliding treatment measures. Landslide kinematics are characterized by subzone-independent displacements and their drivers，which are highly correlated with hydrometeorological extremes. The RTS-based early warning model provides specific hydrometeorological thresholds，emphasizing the emergency response-oriented landslide monitoring and early warning concept.

In order to systematically explore the root causes of industrial hot work accidents through a large amount of text data，a text mining method based on 24Model was proposed. Firstly，220 hot work accident reports were collected and sorted as datasets，and a 24Model classifier based on Bidirectional Encoder Representations from Transformers (BERT) was constructed. The pre-trained model was used to train and evaluate the accident report dataset to construct a classification model. Then，through the combination weight of the Keyword extraction algorithm based on BERT (KeyBERT) and Term Frequency-Inverse Document Frequency (TF-IDF) algorithms，combined with the 24Model framework，a keyword index system for hot work accident text was established. Finally，the interrelationships between accident causes were obtained through the analysis of the network co-occurrence relationship between text-mining keywords. The results show that the BERT-based 24Model classifier model can systematically and accurately determine the causative categories of hot work accidents. The weight of the safety management system was the largest among the 4-level keyword index systems obtained through the combination of weights. Furthermore，7 key causative factors of hot work accidents were obtained by combining them with the co-occurrence network analysis. This shows that 24Model can strengthen the interpretability of text mining results，which provides an important reference for the prevention and management of hot work accidents.

To address the safety risks caused by rapid fire spread and thermal smoke accumulation in confined spaces such as basements and underground warehouses due to lithium-ion electric bicycle fires，this study combines experimental and numerical simulation methods to investigate the temperature field evolution，flame characteristics，and smoke diffusion behavior of lithium-ion electric bicycle combustion under different ambient temperatures，clarifying the effect of temperature on combustion characteristics. The results show that under experimental conditions，the average temperature at the combustion center in the confined space is approximately 600 ℃，with a peak temperature exceeding 920 ℃. Ambient temperature significantly affects the initial stage of lithium-ion electric bicycle combustion. In a 40 ℃ environment，the time for combustion to enter the rapid development phase is shortened by 20 seconds compared to 20 and 0 ℃ environments. The time required for nearby temperatures to reach the ignition point of the lithium-ion electric bicycle is reduced by approximately 25 seconds compared to 20 and 0 ℃ environments. After about 80 seconds，the temperature rise rates converge. At 20 seconds，the flame morphology of lithium-ion electric bicycle combustion differs noticeably across environments：the flame height at 40 ℃ is approximately 1.15 times that at 20 ℃ and 1.32 times that at 0 ℃. Flame morphology converges after about 80 seconds. Within the first 30 seconds，the smoke diffusion velocity and production rate in a 40 ℃ environment are significantly higher than those at 20 and 0 ℃，but smoke concentrations stabilize to similar levels after approximately 50 seconds.

In order to solve the problem that the traditional spray technology in coal mine dust pollution control is not ideal and the spray dust reduction performance is low，Sodium dodecyl sulfate (SDS)，Coconut Diethanol Amide (CDEA) and Cocoamidopropyl Betaine(CAB-35) were selected in this paper. The wettability of the three surfactants was first analyzed macroscopically by contact angle test. Then，combined with molecular dynamics simulation and quantum chemical analysis，the action mechanism of single/compound surfactants on bituminous coal was studied from a microscopic perspective. The results show that among the single surfactants，SDS has the smallest contact angle and the largest molecular orbital energy difference，and is more likely to form high-strength hydrogen bonds with water molecules； the mixed surfactants all show better wettability of bituminous coal than single surfactants，and the contact angle is smaller when non-ionic surfactants are mixed with anionic or zwitterionic surfactants，and the contact angle reduction rate is also much greater than that of single solutions，and the synergistic effect is more prominent; In the CDEA+CAB-35 (4∶2) system，water and surfactants form more and higher strength hydrogen bonds，and the surfactant molecules pull each other to form a tight adsorption layer，which attracts water molecules to infiltrate the surface of coal dust and improves the wettability of coal dust to the best extent.

In order to improve and advance the nuclear radiation detection and monitoring integrated technology based on pixel sensors，a radiation noise suppression and nuclear detection method leveraging the parallel advantages of FPGA was proposed，with corresponding programs developed. By analyzing the characteristics of radiation noise signals，radiation noise suppression and two-dimensional wavelet transform programs based on FPGA were developed to output clear radiation field images. The images were decomposed into horizontal，vertical，and diagonal components，and the results of linear fitting statistics for each component were investigated to identify the component with the best linear fit. The research results demonstrate that the FPGA program modules effectively execute radiation noise suppression and nuclear detection functions in images. After noise reduction，the peak signal-to-noise ratio(PSNR) of the images is improved by approximately 11 dB. The diagonal component is shown to best characterize the radiation response information of the images，achieving a linearity of 0.99624 in linear fitting for different dose rates.

In order to predict the slope state more accurately and effectively prevent the slope instability accident，an improved ISSA-KELM slope stability prediction model was proposed. Firstly，six main factors such as bulk density and cohesion in slope instability characteristics were used as prediction indexes to establish a data set for slope stability evaluation. Secondly，SSA was enhanced by incorporating Sine chaotic mapping，Levy flight strategy，dynamic adaptive weights，and fusion of optimal explosion strategy and reverse learning. These improvements aimed at enhancing the global search capability and stability of SSA. Subsequently，ISSA was employed to optimize the kernel parameter ψ and regularization coefficient C in KELM for improved prediction accuracy while avoiding overfitting issues associated with KELM. The results show that the accuracy rate，precision，recall rate and F₁ score of ISSA-KELM model reached 0.945 9，1，0.866 7 and 0.929，respectively，which are superior to SSA-KELM，PSO-KELM and PSO-SVM models，and the predicted results of the model are the closest to the actual values. It shows that the established ISSA-KELM model has strong generalization ability.