To solve the problem that construction safety requirement information hidden in project documents is hard to be discovered without relevance and semantic ambiguity, a two-stage integration framework combining NLP techniques was developed for project document analysis and classification and extraction of requirement information. First, the safety targets of the project to be evaluated were obtained by combining the multivariate techniques of NLP, and an association model was established based on the topic model to recommend the appropriate requirement types. Then, the semantic features of the three types of elements were considered, and keyword analysis, sentiment analysis, and dependency analysis were adopted to extract the three types of elements, respectively. Finally, two types of construction projects (civil and industrial) were used as case to test the type recommendation and extraction of construction safety requirements. The results show that the two-stage integration framework recommends four appropriate requirement types for civil and industrial buildings respectively, and the combination of lexical properties and lexical sentiment can effectively extract the requirement keywords and behavior opinion words, and the extraction accuracy of the main elements can reach 88.6% after supplementing the description of building types. The test results confirm that responding to safety target features can recommend suitable types from the complicated requirement information, and the classification and extraction of requirement information combined with NLP avoids subjective preferences and improves the accuracy of information extraction.

In order to enhance the application of double-row inclined-vertical pile composite support in loess foundation pits, model tests were conducted to investigate the displacement, internal force, and earth pressure changes of forward-inclined rear-vertical double-row piles and double-row inward-inclined pile support structures during excavation. Numerical analysis was also used to explore the effect of pile inclination on support performance. The results show that the horizontal displacement of the pile body and surface settlement outside the pit are both smaller in the inward-inclined double-row pile support compared to the forward-inclined rear-vertical pile support. Additionally, the maximum axial force, bending moment, and net earth pressure on the front row piles are also lower. As the pile inclination increases, the pile top displacement, surface settlement outside the pit, and the maximum axial force of the pile body decrease, while the maximum bending moment and the safety factor of the foundation pit increase. In the double-row inclined-vertical pile support structure, the front inclined piles function to brace internally, retain the soil, and enhance structural safety and stability, whereas the rear vertical piles serve to anchor, transfer, and distribute earth pressure. Together, they improve the overall stability of the support system. During excavation, the stress path of the soil outside the pit initially moves away from and then approaches the failure K_f line. A greater pile inclination results in the stress path staying further from the K_f line. Comparatively, the stress path of the inward-inclined double-row pile support stays further from the principal stress K_f line, leading to higher safety and stability of the foundation pit.

To address the challenges of high risk and difficulty in quantitative analyzing hazardous chemical storage tank area leakage accidents, a composite method based on STAMP was proposed to elucidate the accident mechanisms. It clarifies the logical relationships among causal factors and quantitatively evaluate the impacts of accident causation, thereby enabling nonlinear quantitative accident analysis. First, the STAMP-24Model was utilized to construct an accident analysis diagram for hazardous chemical storage tank leakage, identifying system components, hierarchical relationships, as well as analyzing accident causal factors and their logical connections. Subsequently, Interpretive Structural Modeling (ISM) method was applied to determine path relationships and hierarchical structures among causal factors. Node importance analysis based on degree and clustering coefficients, as well as BN node analysis, was conducted to assess the criticality of causal factors on system. Finally, the validity and feasibility of the method were verified through a case study. The results show that organizational management failure (e.g., failure to implement safety rules and regulations, lax implementation of engineering management regulations) is the core driver of accident risk evolution, with a total degree value of 53.3%, and a significant coupling effect with physical and personnel layer factors.

With the development of hydrogen-methane-mixture, mixed gas explosion accidents have gradually attracted attention. To reduce the loss and influence caused by mixed gas explosion accidents, this study integrated research results at home and abroad through literature review. The experimental data and numerical simulation analysis results were comprehensively sorted out. The research progress at home and overseas on the influence of initial temperature, initial pressure, gas composition, test vessel and ignition conditions on the explosion limit and explosion pressure of mixed gas was discussed. The results show that the initial temperature and initial pressure have a significant effect on the explosion limit of the mixed gas, and the coupling effect of the two is greater. The upper and lower explosion limits of the mixed gas exhibit different sensitivities to the influencing factors, with the upper limit being more significantly affected. The maximum explosion pressure of the mixed gas generally appears when it is slightly higher than the theoretical stoichiometric concentration, and the influencing factors are complex and diverse. Among these factors, the maximum explosion pressure is greatly affected by the composition of the mixed gas. The addition of combustible gases such as hydrogen and ethylene will increase the explosion pressure of major gases such as methane. Numerical simulation technology can provide an important supplement and verification for the study of explosion limit and explosion pressure of mixed gas.

To identify the primary types and distribution patterns of risky driving behaviors at the entrance and exit of the highway tunnel, a risk driving behavior spectrum was constructed based on trajectory data. Firstly, trajectory data were collected through continuous video surveillance of Liupan Mountain Tunnel in Ningxia. Vehicle traffic characteristics at tunnel entrance and exit were analyzed and four categories of risky driving behaviors were selected, including rapid speed change, serpentine driving, dangerous following, and dangerous lane changing. Then, the measurement of risk method was applied to quantify four types of risky driving behaviors, and quartile deviation and criteria importance though intercriteria correlation (CRITIC) were used to determine the threshold values of risky driving behavior characteristics and weights, and the characteristic values of drivers' risky driving behavior spectrum were calculated. Finally, driving behavior scores were compared with characteristic thresholds, and the spatial distribution of hazard points for typical risky behaviors of dangerous drivers was statistically analyzed. The results of the study show that high-risk drivers in the entrance/exit sections of the tunnel mainly show the behaviors of rapid speed change or serpentine driving, in which the rapid speed change behavior has the most significant effect on the total score of the risky behavioral spectrum of the drivers. Vehicles have large speed variations at both the inside the tunnel entrance and outside the exit from 0 to 50 meters. Moreover, larger speed changes are observed at the exit section (0-50 m) compared to the entrance section, and risk points for rapid speed change behavior of dangerous drivers are also more concentrated in that segment. The lateral offset values outside the tunnel entrance/exit are larger than those inside the tunnel, and the risk points for serpentine driving behavior of dangerous drivers are mainly distributed in the range of 0-50 meters outside the tunnel entrance and exit. Risky driving behavior spectrum based on trajectory data can assess and quantify the risk level of drivers in the tunnel entrance/exit sections, facilitating precise identification of high-risk drivers.

To enhance the identification and control of accident risks in confined space operations in industrial and commercial enterprises, a socially acceptable risk standard for such operation accidents was established, and a risk grading study based on F-N curve was conducted. First, historical data on significant and higher-level accidents in confined space operations from 2014 to 2021 were analyzed statistically in terms of fatalities, accident frequency, and the number of enterprises involved, in order to identify accident patterns. Then, considering accident frequency and fatality consequences, a socially acceptable risk standard was developed based on ALARP criteria, and F-N curve was plotted. Finally, a risk grading management model was established based on three risk zones-intolerable, acceptable, and negligible. The control measures were proposed from both enterprise and individual perspectives. The results show that for accidents with 3 to 10 fatalities, the frequency falls within the magnitude of 10^-6-10^-4. The F-N curve with a slope of -4.033 is used to divide three distinct risk levels, helping to clarify the boundaries of risk levels and highlight management priorities.

To accurately identify deep foundation pit collapse risk levels and improve construction safety, a risk assessment model was developed using variable fuzzy set theory. Sixteen influencing factors were screened from four categories: hydrogeology, support conditions, construction operations, and management monitoring. A risk index system was constructed based on these factors with clearly defined grade standards for risk classification. The multiplicative synthesis method was applied to optimally combine subjective weights derived from Stepwise Weight Assessment Ratio Analysis (SWARA) and objective weights calculated via the entropy weight method, forming comprehensive weights that integrate expert judgment and data objectivity. Variable fuzzy set theory was then utilized to process sample data, generating comprehensive grade characteristic values to determine collapse risk levels. Engineering case results show that the evaluation outcomes of method align with actual construction conditions, confirming its scientific validity and effectiveness. Compared to existing approaches, it more accurately reflects risk status by addressing the fuzzy nature of risk boundaries in deep foundation pit projects. This study provides a scientific and practical framework for risk assessment, enhancing safety management in deep foundation pit construction and offering practical value for ensuring project safety and informed risk management decisions.

To deeply understand the true triaxial mechanical characteristics of coal-rock combinations under mining disturbance, the octahedral shear stress theory was adopted to systematically study the true triaxial mechanical characteristics and energy evolution mechanism of coal-rock combinations affected by the interface inclination angle during mining disturbance. Finally, based on Drucker-Prager criterion, a true triaxial damage constitutive model for coal-rock combinations considering the influence of the interface inclination angle was established. The results show that during the excavation disturbance stage, the axial strain of the coal-rock combination increases linearly with time. The energy is mainly converted into elastic energy, and the dissipative energy gradually increases as the strain increases. When the interface inclination angle exceeds 60°, the shear stress increases significantly, and the node of shear stress increase is postponed to the end of the excavation stage. During the mining stage, the energy of the specimen is mainly used for deformation and failure. When the interface inclination angle is less than 60°, the specimen undergoes two deformations before the shear stress reaches the strength limit. At the end of the mining disturbance, there are two sudden increase nodes in strain, and then the shear stress drops sharply, with the proportion of dissipative energy approaching 1. When the interface inclination angle is greater than 60°, the peak value of axial strain decreases, the sudden increase node of strain is postponed and is synchronized with the sudden drop node of shear stress, and the deformation and failure of the combination mainly depend on the rock mass part. After model verification, the goodness of fit is higher than 90%.

In order to investigate the performance of the latest Chinese Planned Controlled Access Highway Network (CAH-NET), A modeling method in Space-P was proposed, a composite importance index for nodes was introduced, and a vulnerability analysis approach based on node failures guided by this index was presented in this study. Based on the data from the 2022 National Highway Network Planning, the CAH-NET model was constructed, and its static topological characteristics as well as network vulnerability under various attack strategies were analyzed. Results show that the current CAH-NET exhibits typical small-world and scale-free properties, with good overall accessibility. The top rankings for node degree, centrality, and composite importance are mostly occupied by cities in the eastern coastal areas or key central hub cities. The vulnerability of network is slightly affected by random attacks. About 90% of nodes t need to be attacked to completely destroy the CAH-NET. Among intentional attack strategies, the connectivity of network declines most quickly. When the proportion of nodes attacked through the betweenness centrality attack strategy reaches about 30%, the network totally collapses.

To rigorously assess the complexity of operator task execution under nuclear power plant accident scenarios, this study proposes an operator task-handling complexity (OTHC) evaluation model based on information entropy and image entropy. The model evaluates complexity from three dimensions: interface information complexity (IIC), task complexity (TC), and protocol complexity (PC). It establishes an evaluation system comprising eight complexity indicators, including interface layout, icons, text, and color. The entropy weighting method was employed to objectively assign weights, enabling quantitative evaluation of complexity using the model. To validate the model's effectiveness, a primary circuit small-break loss-of-coolant accident diagnosis task was used as a case study, for which a complexity network diagram is constructed, and the proposed model is applied to assess complexity. Subjective workload data are collected using the national aeronautics and space administration task load index (NASA-TLX) scale. The results show that the OTHC of the primary circuit is higher than that of the secondary circuit. The complexity assessed by the constructed model is highly positively correlated with subjective complexity, effectively reflecting the actual level of task processing complexity.