[Objective] Frequent global flooding has led to increasing challenges for urban drainage systems, including not only extreme rainfall, but also the uncertain threats posed by potential structural failures such as pipe deterioration and bursting. In order to study the level of resilience of urban drainage systems under extreme external rainfall and internal structural failure, [Methods] considering the influence of pipe age and flow velocity, the structural failure probability calculation method is proposed, and the overloading scenario without considering structural factors and the composite scenario with considering structural factors are constructed. Based on the comprehensive consideration of the preparatory stage, response stage and recovery stage of the response process of urban drainage system to extreme loads, the resilience assessment indexes are selected and a resilience assessment model based on cloud model is established. The proposed method is applied to the study area for case study analysis. [Results] The result show that the total flood volume in the urban drainage system under the composite scenario increased by more than 60% compared with the exceeding load scenario, and the average flood duration increased by more than 1.0 h. After topology optimization, the total flood volume in the urban drainage system under the exceeding load scenario and the composite scenario decreased by 24.7%~30.1% and 20.5%~48.4%, respectively. [Conclusion] Structural failure has a certain impact on the resilience of urban drainage systems, leading to a decrease in drainage capacity; topology optimization can improve the system resilience to a certain extent. The result show that the proposed method can effectively assess the resilience of urban drainage systems under extreme loads and provide a reference for the safe operation of urban drainage systems under extreme loads.

[Objective] Global warming has intensified the frequency of extreme events such as floods and heat waves, leading to an increased risk of successive flood-heat extreme(SFHE) events. Studying the spatiotemporal evolution of SFHE events has important scientific significance for the prevention and response to composite disaster risks. [Methods] Based on observation and reanalysis data, multiple statistical analysis method were used to analyze the spatiotemporal evolution of SFHE characteristics(frequency, duration, interval between floods and heat waves IFH, coverage area) in nine major river basins in China over the past 54 years(1961—2014), and explore the relative impact of floods and heatwaves on SFHE occurrence. [Results] The result indicate that:(1)The frequency, duration, and land coverage area of SFHE show a significant increasing trend, while IFH has a significant decreasing trend in China.(2)Spatially, the risk of SFHE is higher in the middle of the Southwest River basin, the south of the Yangtze River basin and the north of the Pearl River basin(>34 times/decade), while the duration is longer in the lower reaches of the Yangtze River basin, the Southeast River basin and the south of the Huaihe River basin(>13 d).(3)The periodicity of SFHE frequency varies in different watersheds, but there is a main cycle of 2.8a in China, which is most significant from the 1990 s to the early 2000 s.(4)The frequency of SFHE occurrence in the nine river basins is dominated by flood frequency, with the greatest impact in the Huaihe River basin. [Conclusion] The research result demonstrate that the risk of SFHE occurrence is significantly increasing, and the suddenness of SFHE is increasing in China. Overall, floods play a dominant role in the occurrence of SFHE.

[Objective] Under the joint influence of various factors such as climate change and urbanization, urban waterlogging disasters occur frequently worldwide, the role of urban blue-green infrastructure in source emission reduction and emergency response to excessive runoff regulation is further highlighted. It is necessary to strengthen the synergy of blue-green-gray infrastructure for urban waterlogging regulation. [Methods] Based on conceptual analysis and water balance, the basic synergy of urban waterlogging control concept of the synergy of blue-green-gray infrastructure was systematically elaborated, typical modes of urban rainwater system construction were sorted out, the key technologies of urban waterlogging control of the synergy of blue-green-gray infrastructure were analyzed, the challenges faced by current urban waterlogging control of the synergy of blue-green-gray infrastructure were analyzed, and the development trend was outlooked. [Results] The concept and typical modes of urban waterlogging control based on blue-green-gray infrastructure synergy were systematically expounded, the typical modes of urban waterlogging control based on blue-green-gray infrastructure synergy were emphatically analyzed, and the basic characteristics, main application areas and conceptual diagrams of the typical model of urban waterlogging control with blue-green-gray infrastructure synergy were analyzed. [Conclusion] Strengthening the synergy of blue-green-gray infrastructure is an important development direction of the urban waterlogging control. In the future, it is necessary to further strengthen the overall planning of blue-green-gray infrastructure, facility optimization and intelligent management and control, and promote the urban rainwater system to expand in the direction of synergy, resilience, low carbon and wisdom.

[Objective] Flash floods are often exacerbated by factors such as flood diversion, sediment deposition, and debris blocking bridge openings, Resultsing in significant uncertainty. Using a hydrologic-hydrodynamic model, the amplification effects of bridges under different levels of blockage on flash flood disasters was quantitatively assessed. [Methods] Taking the Zhaigang River in Liannan County, Qingyuan City, Guangdong Province as an example, the China Flash Flood Hydrological Model(CNFF-HMS) as hydrological model and the Integrated Flood Modeling System(IFMS) as hydrodynamic model were employed. Parameter calibration and result validation of the hydrological and hydrodynamic models were conducted using 11 and 4 sets of rainstorm flood data, respectively. Using the typical flood event on June 21, 2022, variations in water levels and inundation extent at the bridge cross section were calculated under three conditions: unblocked, partially blocked, and fully blocked. Contributing factors of water blockage caused by bridges were also analyzed. [Results] The results showed that:(1) bridge blockage caused a significant rise in upstream water level, with a 1.5-meter increase in water level under the fully blocked condition compared to the unblocked condition.(2) Bridge blockage led to a noticeable expansion in the inundation extent. Specifically, the inundation area under the fully blocked condition was 0.99 km² larger, or 1.34 times larger than the unblocked condition.(3) Bridge blockage reduced the bridge's conveyance capacity, with the flow area under the fully blocked condition being only 38.4% of that under the unblocked condition. [Conclusion] Bridge blockage has an amplification effect on flash flood disasters. As the severity of bridge blockage increases, the bridge's conveyance capacity weakens, and the upstream water level rises more noticeably. This significantly increases the risk of flooding and expands the inundation extent. In this study, the amplification effect of the disaster has been quantitatively assessed, providing more precise and scientifically grounded support for the early detection of flash flood disaster risks and for guiding evacuation and relocation efforts.

[Objective] To explore the spatial and temporal variation characteristics of vegetation ecological quality in Jinhua City in the past 20 years, and to analyze the driving factors of its spatial distribution, in order to provide important theoretical support for the study of dynamic changes of vegetation ecosystem and the protection of vegetation resources in Jinhua City. [Methods] Based on meteorological satellite remote sensing data, the vegetation ecological quality of Jinhua City in the past 20 years was analyzed and studied by using linear regression and geographical detector. [Results] The vegetation ecological quality in Jinhua City has continued to improve over the past 20 years. The vegetation coverage has significantly increased, with the average vegetation coverage increasing from 56.3% in 2003 to 65.4% in 2022; The net primary productivity of vegetation has increased by an average of 4.9 gC·m^-2 annually; More than 90% of the areas have shown continuous improvement in the Vegetation Ecological Quality Index(VEQI). The four factors of land use, elevation, landform type, and nighttime lighting have a significant impact on the spatial differentiation of VEQI in Jinhua City. The single-factor explanation of night lighting in 2020 is the strongest, and its interaction with natural environmental factors has a stronger explanatory power. Meanwhile, there are significant differences between night lighting and all other factors. [Conclusion] The interaction between climate, surface, and human activities affects vegetation growth in a nonlinear manner, and human activities have a significant impact on the spatial differentiation pattern of VEQI.

[Objective] To explore scheduling strategies for ensuring minimum discharge flow in flow-reduced reaches of the Great Bend of the Yalong River, thereby safeguarding aquatic ecosystem health. [Methods] The Mann-Kendall(M-K) test was used to detect non-stationarity in runoff series. A joint probability distribution model was established for mainstream and tributary inflows of the Luning cross-section of the Yalong River using the Copula function to analyze the wet-dry encounter characteristics of mainstream and tributary runoff. Subsequently, a strategy for ensuring minimum discharge flow at the Luning cross-section under inconsistent runoff conditions was proposed. A joint scheduling model for cascade reservoirs was established to analyze schemes for ensuring minimum discharge flow at the Luning cross-section and scheduling strategies for cascade hydropower stations under normal and extremely dry scenarios in the mainstream and tributaries. [Results] The result showed that an abrupt change in runoff occurred at the Jinping-II dam site around 1989. Both pre-and post-change periods exhibited low probabilities of wet-dry or dry-wet encounters in mainstream and tributary inflows. Compensation water releases from the Jinping cascade were necessary to maintain minimum discharge flow at the Luning cross-section. Under normal and dry scenarios of mainstream and tributary inflows, Jinping-II gate site required additional compensation water releases of 0 to 18.2 m³/s and 11 to 48 m³/s, respectively, beyond the ecological base flow. During intraday no-load operation of the Jinping cascade, the hydraulic connectivity between upstream and downstream reservoirs was leveraged to enhance the regulation capability of Jinping II. Joint scheduling of mainstream and tributaries was implemented to achieve the goal of minimum discharge flow at the Luning cross-section. [Conclusion] The M-K method effectively identifies abrupt change points in runoff at the Jinping-II dam site. The joint probability distribution model based on Copula function accurately describes the dependency between mainstream and tributary runoff. The proposed scheme for ensuring minimum discharge flow and scheduling schemes for cascade hydropower stations under different scenarios can effectively ensure the minimum discharge flow in flow-reduced reaches of the Great Bend of the Yalong River, thereby preserving aquatic ecosystem health.

[Objective] The operation of hydropower plants not only alters the suitability of downstream river habitats through changes in discharge but also affects the habitat characteristics of the fluctuating backwater area at the reservoir tail due to water level fluctuations in front of the dam. By clarifying the effects of hydropower plant operation on the suitability of fish spawning habitats in the backwater area under different water inflow conditions, this research aims to provide a reference and guidance for ecological scheduling during fish spawning periods in regulating hydropower plants. [Methods] Using the operational data from the Longyangxia hydropower plant in the upper Yellow River as a case study, this research focused on the Yehuxia spawning ground at the reservoir tail as a fish protection habitat. Platyharodon extremus was selected as the target species. A fish habitat model was constructed, and the Weighted Usable Area(WUA) and the Splitting Index(SPLIT) were introduced to quantitatively assess how changes in hydropower plant scheduling affect habitat suitability during fish spawning periods. [Results] The result indicated that the current annual operational mode affected the fish habitat at the reservoir tail, with the WUA first increasing and then decreasing, while the SPLIT of the spawning habitat gradually increased. From early April to late May, the WUA increased by 74.43%. From late May to late June, the WUA decreased by 52.39%. From April to June, the SPLIT increased by 221.94%. The inconsistent trends of WUA and SPLIT were due to the fact that WUA reflected changes in the spawning habitat area, while SPLIT represented the integrity and fragmentation of the spawning area. From early April to late May, increases in flow and decreases in water level expanded the spawning area. However, changes in topography and flow patterns worsened the overall integrity, causing both WUA and SPLIT to increase. From late May to late June, as flow levels continued to rise and the reservoir water level increased, the suitable spawning areas shrank and became more fragmented, leading to a decrease in WUA and an increase in SPLIT. [Conclusion] The result indicate that the annual operational mode at Longyangxia significantly affects both SPLIT and WUA. Further ecological scheduling of the hydropower plant can effectively improve the ecological hydrological processes and hydrodynamic conditions of fish spawning habitats in the reservoir tail backwater area.

[Objective] In the context of global climate change, rapid urbanization has a profound impact on hydrological processes, and complex urban underlying surface has brought challenges to the calculation of runoff generation and flow concentration. There is an urgent need to develop urban flood simulation technologies that comprehensively consider different underlying surface conditions. [Methods] Based on the characteristics of runoff generation and flow concentration and land use types, the complex urban underlying surface was classified into permeable, directly impervious, and indirectly impervious surfaces. The runoff generation was calculated using the saturation-excess, infiltration-excess, and API-SCS-CN method, respectively. Combined with the flow concentration of slopes, pipelines, and river channels, Chebeichong urban flood simulation model was established for the Chebeichong River Basin in Tianhe District, Guangzhou. The model's practicality, considering the complex underlying surface, was validated using the measured rainfall and runoff data from 2021 to 2023. [Results] During the calibration period, the Chebeichong urban flood simulation model achieved a Nash efficiency coefficient of 0.71 and a correlation coefficient of 0.74 for flow simulation. During the validation period, the Nash efficiency coefficient reached 0.79, and the correlation coefficient reached 0.81. Under the same modeling conditions, the SWMM-SCS model achieved a Nash efficiency coefficient of 0.5 and a correlation coefficient of 0.6. [Conclusion] The simulation result of the Chebeichong urban flood simulation model are more accurate than those of the SWMM-SCS model, with the Nash efficiency coefficient improving by 66.7%, the runoff correlation coefficient increasing by 29.7%, the flood volume error decreasing by 48.3%, and the peak flow error decreasing by 3.2%. These findings indicate that incorporating complex underlying surfaces and multiple runoff generation models enhances the reliability and applicability of urban flood models, providing a new approach for urban flood simulation.

[Objective] With the increasing importance of urban drainage pipelines in flood management and water pollution control, there is an urgent need for optimized monitoring networks. [Methods] An efficient and accurate method for optimizing the layout of monitoring points in urban drainage pipelines to detect and identify illegal sewage discharge was proposec. A Storm Water Management Model(SWMM) was utilized to simulate pollutant transport within urban drainage pipelines and generate time-series data. An agglomerative hierarchical clustering algorithm was applied to classify the data and determine the optimal number of monitoring points in the drainage network. Node correlation was assessed through cross-correlation functions, selecting nodes with the highest coefficients to ensure comprehensive pollution conditions within the clusters. [Results] The proposed method was validated using a drainage network case from the SWMM manual, and it was determined that deploying three monitoring points is the optimal choice. The evaluation revealed a reliability of 91.86% and an average response time of 3.26 minutes at three monitoring points. Agglomerative hierarchical clustering outperformed the K-means algorithm in terms of monitoring point layout effectiveness, especially in covering the upstream, midstream, and downstream sections of the drainage network. [Conclusion] This method offers a new technical means for optimizing the layout of monitoring points in urban drainage pipelines, enhancing urban drainage monitoring and management and offering new perspectivesfor related research areas.

[Objective] With the intensification of global climate change, the frequency and intensity of extreme rainfall events have significantly increased, often overwhelming traditional urban stormwater drainage systems. To mitigate the risks of urban flooding under changing climate conditions, a climate adaptation optimization method for low-impact development(LID) stormwater drainage systems was proposed, based on projections from the CMIP6(Coupled Model Intercomparison Project Phase 6) models. [Methods] Taking a district in Jiujiang City as a case study, the optimal CMIP6 rainfall projection scenario was selected by integrating Taylor diagram scores and interannual variability scores from various models. These rainfall scenarios were then used for multi-objective optimization of the SWMM(Storm Water Management Model) with NSGA-II(Non-dominated Sorting Genetic Algorithm II). The scale of the LID facilities was optimized based on Pareto optimal result. [Results] The findings indicated that all three models predict increased rainfall under various future climate scenarios, with FGOALS-g3 simulating the highest total annual rainfall and daily rainfall, along with greater uncertainty. As investment in LID facilities increased, both runoff volume and the number of overflow nodes were effectively controlled. The moderate investment scenario, costing 2.602 7 million yuan, reduced runoff by 21.55%, while the maximum investment scenario, costing 5.195 8 million yuan, reduced runoff by 25.00%. The number of overflow nodes decreased by 14.12% and 18.82%, respectively. Compared to the baseline conditions, the peak runoff under the moderate and maximum investment scenarios was reduced to 0.64 m³/s and 0.62 m³/s, accounting for approximately 86.49% and 83.71% of the original values, respectively. [Conclusion] The result demonstrate that future climate change will place greater pressure on urban flooding due to extreme rainfall events. Analysis of the simulation result indicates that both the moderate and maximum investment LID facility plans effectively control runoff volume and the number of overflow nodes. This suggests that the optimization strategy is effective in addressing future extreme rainfall impacts. The method provides a valuable reference for flood risk management and adaptive urban planning in similar built-up areas.

[Objective] The optimization and allocation of water resources was proven to be crucial in enhancing the overall efficiency of water utilization, aiming to maximize and optimize the overall benefits of water resource utilization. [Methods] Facing the challenge of formulating water allocation plans for reservoirs with irrigation tasks in irrigation districts, a water resources optimization and allocation model was proposed, targeting maximized comprehensive benefits during the reservoir operation period, maximized storage benefits at the end of the period, and minimized water discards. A benefit function for end-of-period storage was constructed, and the multi-objective problem was transformed into a single-objective model, which was then solved using the genetic algorithm. Taking the Zhanghe Reservoir Irrigation District in Hubei Province as a past example, the model was utilized, based on a water demand forecasting model, to analyze domestic, industrial, and irrigation water demands. Considering the uncertainty of inflow at the end of the reservoir operation period, a relationship for end-of-period storage benefits was established. The model then facilitated the optimal allocation of water resources. [Results] The result revealed that the proposed method had yielded higher overall benefits. Based on water demand predictions for the year 2024 and under different rainfall frequencies(P=25%, 50%, 75%), compared to the reservoir operation chart method that had been used, the comprehensive benefits of the reservoir were increased by 610,000 yuan, 300,000 yuan, and 130,000 yuan, respectively, under different rainfall frequencies of P=25%, 50%, and 75%. Meanwhile, the end-of-period water levels were lowered by 1.9 m, 0.85 m, and raised by 0.25 m, respectively. [Conclusion] The quantification of end-of-period storage benefits, which represent the risk of comprehensive benefits arising from uncertainty in future inflows after the operation period, played a significant role in water resources optimization. This approach enabled the optimal allocation of water resources during the reservoir operation period to achieve maximum economic benefits in the reservoir-irrigation district system. It also guided the setting of end-of-period water levels, thereby enhancing the overall efficiency of water utilization from the water source perspective.

[Objective] Radial gates are crucial components in flood discharge structures for reservoirs and dams. Their structural safety and reliability are of great significance for achieving normal operation of reservoirs and dams and ensuring structural safety during flood seasons. To analyze the structural safety of the radial steel gate under multiple complex load conditions, a detailed 3D numerical model of the large-size radial steel gate was developed using ANSYS. [Methods] Taking the large-size radial steel gate in the Kafue Gorge Lower Hydropower Station as an example, the mechanical responses of the radial gate under three typical operating cases, e.g., closed, instant opening, and momentary open fault, were calculated. Then, the reasonableness of the 3D model and simulation process was verified through case comparisons. Moreover, the distributions and variation laws of displacement and stress of the radial gate under different operating conditions were further analyzed. The strength, stiffness, and stability of the main structure and fastening components of the radial gate were analyzed comprehensively. [Results] The result show that the case of momentary open fault constitutes the most critical operating condition for the structural safety of the main structure and the radial gate. Under this scenario, significant deformation and high concentration of equivalent stress are observed around the connection between the lower support arm and the gate leaf on the opening and closing force application side, which is identified as the dangerous part. Overall, the main structure and anchoring components of the radial gate meet the relevant requirements for strength, stiffness, and stability. [Conclusion] The results indicate that in the long-term service state of the radial gate, the dangerous part is prone to wear and has a risk of fatigue damage due to repeated opening and closing cycles. Comprehensive analysis suggests optimizing the design and reinforcing this structural area to meet the safety requirements under the most unfavorable conditions, thereby ensuring the long-term safety of the structure. The structural design of the radial gate is safe and reliable. This study ensures the structural safety of the large-size radial steel gate in the Kafue Gorge Lower Hydropower Station and can serve as a reference for the safety design of similar radial gates.

[Objective] The water conservancy project in the north bank irrigation area of Xiaolangdi was taken as the research object, aiming to explore the large deformation phenomenon after excavation and support of local tunnel section, analyze the lateral deformation mechanism of surrounding rock of sand-mudstone interbedded tunnel under heavy rainfall conditions, and put forward reasonable tunnel excavation and support measures. [Methods] By means of field investigation, field monitoring and numerical simulation, the lateral deformation of tunnel surrounding rock, the characteristics of plastic zone and the stress state of supporting unit under different working conditions were studied. [Results] The results show that the maximum displacement of the side wall of the tunnel is relatively small under the initial condition, and the surrounding rock is basically stable. Under the condition of rainfall, the lateral deformation of the tunnel foot is large, and the supporting equipment such as steel arch is destroyed. Under the condition of rainfall and support, the maximum lateral displacement of the side wall can be almost ignored, and the support effect is good. [Conclusion] The lateral deformation of the tunnel is a gradual process. The lateral deformation mechanism of surrounding rock of 9~# tunnel is as follows: the lens of local weak rock layer affects the stability; heavy rainfall infiltration and agricultural irrigation caused the rise of groundwater level, increased pore water pressure and reduced the strength of surrounding rock, [Results] ing in insufficient bearing capacity of the original supporting structure, which could not limit the development of deformation and cause large deformation. Adding transverse support at the bottom can significantly inhibit the lateral deformation of surrounding rock, and the effect is obvious at the bottom feet on both sides. The support optimization scheme of ‘adding transverse steel support at the bottom + grouting reinforcement along 2 m around the hole' is better.

[Objective] Due to the limited fault data during actual operation of hydropower units and differences in fault signal distributions across different units that do not conform to the assumption of identical distribution, model training becomes difficult, and existing diagnostic models have poor generalization capabilities. To address these issues, a fault diagnosis method combining transfer learning strategies with a CNN-BIGRU-attention network(TCBA) is proposed. [Methods] Rotor test bench data was used as the source domain, and real vibration data from hydropower units served as the target domain data. First, a CNN-BIGRU-attention diagnostic model was constructed by combining Convolutional Neural Network(CNN), Bidirectional Gated Recurrent Unit(BIGRU), and attention units. The model was initially trained using source domain data, and its parameters were then transferred to the fault diagnosis model of the target domain. During the transfer process, the lower-layer network was frozen, and the upper-layer network was fine-tuned using partial target domain data, resulting in a fault diagnosis model adapted for the target equipment. To verify the effectiveness of the proposed method, a comparison was conducted between the proposed method and traditional deep learning method through transfer experiments using rotor test bench datasets and real hydropower unit fault data, evaluating indicators such as recognition accuracy, training speed, and sample size requirements. [Results] The result showed that, compared with traditional training method, the proposed method significantly improved the model's convergence speed and effectively reduced the sample size required for training. Under small sample conditions, the fault state recognition accuracy for actual hydropower station fault sample data reached 99.02%, which was about 3% higher than that of the traditional method. [Conclusion] This study demonstrates that the proposed method has strong fault state recognition capabilities, providing an effective solution for fault diagnosis of hydropower units under limited data conditions.

[Objective] In order to address the issue of shutdowns caused by converter protection actions due to inrush currents and transient impulses in high-voltage flexible direct current(HVDC) transmission systems, [Methods] first, the generation mechanism of inrush currents was explained. Then, the impact characteristics of inrush currents on the operation of modular multilevel converter(MMC) under different grid voltages and current collection control method were investigated. Based on this, a transient impulse suppression method for MMC under inrush currents was proposed, which involved switching between the grid-side and valve-side control modes. Finally, an electromagnetic transient model of the actual engineering system was built in MATLAB/Simulink for simulation verification. [Results] The simulation result demonstrated that different collection control method had a significant impact on the transient electrical characteristics of MMC under inrush currents. [Conclusion] Compared to the grid-side control method, the proposed valve-side collection control method significantly reduces the disturbance impact of inrush currents on the converter. The peak current in the converter bridge arm and the voltage fluctuation of the submodule capacitor decrease by 40% and 67%, respectively, which can ensure the safe and stable operation of the flexible HVDC converter.

[Objective] The formation of scour holes is a primary cause of bridge failure under flood conditions. Uncertainty in hydrological data and multi-layer soil erosion significantly impact the accuracy of scour depth predictions. Evaluating the safety performance of bridges over their entire lifecycle under flood-induced scour is essential. [Methods] Using a real bridge project as a case study, the Seasonal Autoregressive Integrated Moving Average(SARIMA) model was applied to conduct a statistical analysis of scour depth over the bridge's lifecycle, based on incomplete hydrological data. Following AASHTO guidelines, four failure modes of bridge pile foundations under flood conditions were considered, and a time-dependent safety performance evaluation model based on probabilistic scour depth was developed. The time-dependent vulnerability curves for the bridge's lifecycle were also obtained. [Results] The result demonstrated that the proposed method effectively predicts the relationship between bridge foundation scour depth and service time, with scour depth initially increasing rapidly before gradually slowing down, particularly over the first 25 years. [Conclusion] The predicted pile foundation scour depth increases with extended service time, and its failure probability is negatively correlated with embedment depth, but is influenced by soil layer parameters. This method provides a theoretical basis for assessing and designing bridge scour over its lifecycle, improving bridge safety under flood conditions.

[Objective] Current pre-site selection of pumped storage power stations heavily relies on manual comparison and selection, which suffers from time-consuming processes and low-automation levels. To address these issues, a method integrating multimodal large models into the pre-site selection of pumped storage power stations is proposed. [Methods] Based on site selection criteria for pumped storage power stations, an evaluation system for potential sites was established. The fuzzy comprehensive evaluation method was employed to calculate an overall score for each site, which served as the station label. Then, specific prompts were designed to guide the GPT model in generating prompt fine-tuning data associated with remote sensing images. Based on this, prompt engineering and Low-Rank Adaptation(LoRA) fine-tuning techniques were used to train the multimodal large language model LLaVA. Subsequently, the trained model was applied to the pre-site selection of the Jixi Pumped Storage Power Station in Anhui Province, followed by a systematic evaluation of the model performance. [Results] The result showed that the model accurately scored for key indicators such as hydrology, topography, and economic factors for the Jixi Pumped Storage Power Station, yielding a comprehensive score of 84.4 that met the criteria for an ideal site. When validated on a test set of 1 091 samples, the model successfully identified 74.1% of ideal site samples and 82.4% of non-ideal site samples. The fine-tuned LLaVA model achieved an Area Under the Curve(AUC) value of 0.822, outperforming Qwen-VL-Chat, InternLM-XComposer-VL, VisualGLM, and InstructBLIP models by 0.106, 0.152, 0.205, and 0.207, respectively. [Conclusion] The findings indicate that the LLaVA model fine-tuned by the proposed method achieves significant improvements in accuracy, recall, and false detection rates for site classification compared to general-purpose multimodal models. Additionally, it demonstrates excellent site evaluation in practical applications, showing high potential for broader application. The domain-specific fine-tuning and application of the LLaVA model effectively highlight the unique advantages of multimodal large models in improving the efficiency and automation level of site selection, providing robust support for the intelligent transformation of the pumped storage industry.

[Objective] Silt from Xuzhou is widely dispersed. In order to comprehend its triaxial compression characteristics and offer a parameter foundation for engineering buildings, [Methods] the silt from Xuzhou Metro Line 1 was subjected to the standard triaxial compression test, in which the axial pressure was applied constantly until the soil was destroyed. The entire test procedure can be thought of as the applied load on the soil's work process. On this premise, the numerical simulation approach is applied to simulate the triaxial test process of Xuzhou silt. [Results] The results show that: With the increase of confining pressure, the peak strength and residual strength of silt increase linearly, and the deformation modulus shows a positive correlation trend; it shows shear failure under different confining pressures, and the rubber membrane distortion is more obvious under low confining pressure. All kinds of cracks gradually increase, the microcracks of the sample begin to increase sharply when the stress reaches the peak value, and the proportion of shear cracks gradually increases; at the initial stage of loading, the load work is almost completely converted into strain energy; during the whole process, the strain energy increases first and then decreases, while the friction energy and damping energy increase until the model is destroyed; the sum of strain energy, friction energy, and damping energy accounts for about 60% of the boundary energy, and the change with the axial strain growth rate is closely related to the stress-strain curve of silt. [Conclusion] The numerical simulation results are in excellent agreement with the laboratory test result. Shear cracks dominate the final failure of silt. According to the energy change, the failure mechanism of the triaxial compression process of silt is separated into three stages: pore compaction, elastic deformation, and plastic deformation before the peak value, and two stages: accelerated failure and strength residual after the peak value. However, with the increase of confining pressure, the post-peak accelerated failure and strength residual stage may disappear.

[Objective] With the increasing depth of mining activities, rockburst has become a major hazard threatening the safety of underground engineering. To quickly assess the rockburst tendency in deep mines, a new evaluation method is proposed. [Methods] The method determines the in-situ stress direction by quantifying the unloading-induced damage characteristics of deep rock cores and measures the stress magnitude using deformation rate analysis(DRA). It further integrates other rock mechanical parameters to evaluate the rockburst tendency in the Jigongling Phosphate Mine in Guizhou Province. [Results] The research results indicate that:(1) the wave velocity and porosity of deep rock cores exhibit consistent orthogonal anisotropy due to unloading damage, thus determining the direction of horizontal principal stress;(2) DRA test result show that within the depth range of 728 m to 1 076.2 m, the minimum horizontal principal stress is 15.5~19.5 MPa, and the maximum horizontal principal stress is 21.1~25.6 MPa;(3) based on the elastic strain energy index(W_et), brittleness index(F), stress intensity ratio(R), and rock mass quality index(RQD) obtained from DRA tests, the rockburst tendency of deep rock masses was evaluated, showing a predominantly moderate tendency. [Conclusion] The proposed method can quickly and accurately measure in-situ stress and key parameters for rockburst evaluation during the early design stages of mine construction, providing important support for assessing rockburst tendency in deep mines.

[Objective] The determination of roof thickness between different sections in underground filling mining of metal mines is very important for stope stability, especially the research on the calculation method of roof safety thickness under the condition of large thickness span ratio and inclined ore bodies is not sufficient. [Methods] Based on the Reissner theory of thick plate(thickness to span ratio >0.125~0.2) and considering the transverse shear deformation of thick plate, the expression of roof deflection under the condition of four sides fixed support is derived. Based on the traditional Janssen's theory of granular pressure, the expression of granular pressure under the inclined Angle is derived for the situation of the roof covered with loose backfill after mining of inclined orebody. Finally, according to the first strength theory, a mechanical model is established to calculate the critical safety thickness of the roof under the inclined backfill. [Results] The result show that the influence of friction Angle on the safety thickness of the roof is more significant than that of the internal friction Angle. The safety thickness is also influenced by the roof size. When the aspect ratio is less than 1.5, the safety thickness increases linearly and rapidly. Between 1.5 and 3.0, the increase rate of safety thickness gradually slows down, and it is a transition stage from rapid growth to stability. When the aspect ratio is greater than 3.0, the safety thickness will become stable. When the dip Angle of the ore body gradually changes from horizontal to vertical, the friction force of the rock side wall on the backfill body decreases, more loads are gradually transferred to the roof, and the reserved safety thickness will increase accordingly. [Conclusion] Through engineering examples, the calculation of roof safety thickness by Reissner theory of thick plate and the extended Janssen's theory of granular pressure can provide a reasonable critical safety thickness of roof after mining and filling of inclined orebody, and the obtained value of roof safety thickness is both economical and safe, the method of determining the safety thickness of underground mining roof in metal mines is further enriched, and the theoretical basis for the safety management of mining roof in similar mines is provided.

[Objective] Due to increasingly severe environmental issues and resource depletion, the sustainable utilization of copper tailings in cementitious materials has attracted widespread attention. However, accurately predicting the compressive strength of cementitious materials incorporating copper tailings remains a challenge due to the complex interactions among material components. A high-precision predictive model was developed using a Stacking ensemble learning approach and optimize mix design to enhance the mechanical properties of the materials. [Methods] Experiments were conducted to investigate the effects of different copper tailings replacement levels(0%, 5%, 10%, 15%, and 20%) and water-to-binder ratios(0.35 and 0.45) on the compressive strength of cementitious materials. To improve the generalization capability of the model, a data fusion method was employed by integrating experimental data with a publicly available concrete compressive strength dataset, [Results] ing in a dataset containing 698 samples. A Stacking ensemble learning model was constructed based on k-nearest neighbors, support vector regression, decision trees, and random forests, with RF serving as the meta-learner. Additionally, Bayesian optimization was applied to fine-tune the hyperparameters of the model to enhance predictive performance. The predictive performance of the Stacking model was evaluated using root mean square error, standard deviation, mean absolute percentage error, and coefficient of determination and was compared with that of individual machine learning models. [Results] The experimental result showed that the compressive strength of cementitious materials generally decreased with increasing copper tailings content, with a significant drop observed when the replacement level exceeded 15%. At the curing age of 28 days, the specimens exhibited the highest compressive strength, indicating a well-developed hydration reaction. The Stacking ensemble learning model demonstrated the best performance in predicting compressive strength, achieving RMSE=0.37, SD=0.16, MAPE=0.91%, and R²=0.991, significantly outperforming individual machine learning models. Among the individual models, RF showed the best performance(RMSE= 2.57, R²=0.977), while KNN exhibited the lowest predictive accuracy(R²=0.967). [Conclusion] A Stacking ensemble learning-based predictive model was developed for the compressive strength of copper tailings cementitious materials and further enhanced its predictive accuracy through Bayesian optimization. The findings indicate that optimizing the water-to-binder ratio and copper tailings content is crucial for improving the mechanical properties of cementitious materials. The proposed Stacking-based predictive model provides reliable data support for mix design optimization, promoting the sustainable application of copper tailings in construction materials.