To improve the prediction accuracy of monthly runoff， this paper proposes a prediction model based on a two-stage decomposition strategy. First， the original runoff sequence is decomposed into trend terms， seasonal terms and residual terms through the Seasonal-trend decomposition using Loess（STL）， and then the residual items with strong randomness are further decomposed through the Variational Mode Decomposition （VMD） to eliminate noise. Three machine learning models， namely， Long-Short Term Memory （LSTM） networks， Convolutional Neural Network （CNN） and Support Vector Regression （SVR）， are used to predict each component one by one， and the monthly runoff prediction result is a linear set of the predicted values of each component. Taking Shimen Station in Lishui Basin as the research object， indexes such as Mean Absolute Error （MAE）、Mean absolute percentage error （MAPE） and Nash Efficiency （NSE） were selected to comprehensively evaluate the model prediction accuracy， and combined with the SHAP（Shapley Additive explanations） interpretable machine learning method to explore the input characteristics in the optimal model contribution to runoff prediction results. The result shows that the overall prediction accuracy of LSTM and CNN is better than that of the SVR model， but the change in prediction accuracy caused by the difference in the model structure is smaller than that caused by the difference in input items； the contribution of each decomposition component in the optimal model STL-VMD-LSTM to the prediction results is better than that of other input characteristics.

The traditional hydrological design in data-starved areas mostly uses hydrological transfer methods， but due to the spatial heterogeneity of the Loess Plateau， the estimation results often have large errors， and the runoff mechanism in the Loess Plateau area is relatively complex， with strong groundwater recharge， so it is difficult for simple hydrological models to accurately simulate the regional hydrological processes. Given this， this paper takes the Malian River Basin as the study area， implements the hydrological simulation on the scale of the Malian River Basin through the SWAT-MODFLOW coupled model， evaluates the spatial and temporal distribution patterns of its surface water resources under different surface cover and its evolution characteristics， to provide a reference for the evaluation of water resources in the data scarce area represented by the Malian River Basin. The model calibration results show that the constructed SWAT-MODFLOW coupled model has good applicability in the loess plateau represented by the Malian River basin， and the NSE coefficient of the outlet site is as high as 0.92. Compared with the area ratio calculation method commonly used in hydrological design， the spatial distribution results of surface water resources obtained based on the coupled model have more reference value. The spatial heterogeneity of runoff and water production capacity of different sub-basins or regions in the Malian River Basin under near-natural conditions is remarkable. The overall trend of runoff at the outlet section is gradually increasing from northwest to southeast， with the strongest water production capacity in the Dongchuan and Xichuan mainstream areas in the upper part of the basin， the weaker capacity in the upper part of the Huanjiang River， and the water production capacity in other areas between the two. The change of vegetation cover in the basin with the transfer from cropland to grassland and woodland has a limited impact on the surface water resources in the Malian River basin， while the direct human activities in the watershed represented by water extraction have a profound impact on the reduction of runoff.

The Jiaodong water diversion project （JWDP） in Shandong Province is an important cross-basin water diversion project to alleviate or solve the water shortage problem in the Jiaodong district of Shandong Province， China， especially in Weifang City， Qingdao City， Yantai City and Weihai City. In order to improve the management level of JWDP and realize fine dispatching operation， the water dispatch optimizations of JWDP based on water supply priority mode and water supply equilibrium mode are carried out， with a limited channel capacity of JWDP. Firstly， an evaluation index system for water supply priority is constructed from three aspects---water shortage degree， water consumption efficiency， and the support of water diversion to social-economic development. The levels of water supply priority of the four cities in Jiaodong District are determined by an entropy weighting method. Secondly， under the same optimization goal （minimize total water shortage）， two water dispatch optimization models are constructed respectively， one based on water supply priority mode and by using the goal planning method， another based on water supply equilibrium mode and by using the linear planning method. Thirdly， the water dispatch optimization models are applied for two situations： scenario 1 （the water demand in 2017-2018） and scenario 2 （a special dry year， the water demand equals diversion index）. The optimized results show that， in the dry year， under water supply equilibrium mode， the total water transfer and the average water transfer satisfaction rate of the four cities are higher than those of the water supply priority operation mode. In the special dry year， under water supply priority mode， only the water demand of the receiving area with the highest priority can be met， and the average water transfer satisfaction rate of the four cities is 61.57%. Under water supply equilibrium mode， although the water demand of the four cities cannot be met， the average water transfer satisfaction rate of the four cities reaches more than 76.81%. Therefore， under the two situations， the average water transfer satisfaction rate of the four cities under the equilibrium mode is higher than the priority mode. This study provides a theoretical support for policymaker of JWDP to choose water dispatching modes， and also a reference for similar water diversion projects.

Influenced by such factors as meteorology， climate change and human activities， the runoff series present unsteady and nonlinear characteristics， which brings a new challenges to the accurate prediction of runoff. In order to improve the prediction accuracy of a single runoff prediction model， a new combined runoff prediction model （EEMD-VMD-SSA-KELM） based on ensemble empirical mode decomposition （EEMD） algorithm， variational modal decomposition （VMD） algorithm and kernel extreme learning machine （KELM） which is optimized by sparrow search algorithm （SSA） is proposed. Firstly， EEMD algorithm is used to decompose the runoff sequence into trend component， detail component and random component， and then VMD algorithm is used to further decompose the random component with the highest frequency into several components with different frequencies that are more stable than the random component to reduce the instability of the runoff sequence. Secondly， the KELM model is established for each component to predict， and SSA is used to optimize the kernel parameters and penalty coefficients of the KELM model. Finally， the prediction results of runoff series are obtained by accumulating the prediction results of all components. The proposed model is applied to the daily runoff prediction in the flood season of Cuntan Hydrological Station in Yichang， Hubei. The proposed model is applied to the flood season daily runoff prediction of Cuntan Hydrometric Station in Yichang， Hubei Province， and is compared with BP neural network model， Least Squares Support Vector Machine Model （LSSVM）， KELM model， etc. The results show that the prediction accuracy of the model combined with data decomposition algorithm is obviously better than that of the single BP model， LSSVM model and KELM model， and the prediction accuracy of the model combined with EEMD algorithm and VMD algorithm was better than that of the model combined with EEMD algorithm only. The prediction accuracy of KELM model is better than that of LSSVM model； The optimization precision of SSA was better than that of particle swarm optimization （PSO） algorithm. The prediction accuracy of EEMD-VMD-SSA-KELM model is the highest， which can accurately simulate the change trend of daily runoff in flood season with complex multi frequency information， and can provide a reference for hydrological forecasting and related forecasting research.

At present， the flood inundation process analysis in a watershed under real-time rainfall cannot meet the needs of practical engineering application because of its complicated implementation procedure. Although the non-source inundation analysis method is highly popularized and is easy to implement， its actual physical significance is not strong. To this end， this paper aims at establishing a cellular automata-based flood inundation analysis model with the aid of real-time rainfall forecast and GIS technologies， in which the GIS technology is used to generate the raster data for digital elevation model. Then， the flood inundation area and water depth can be obtained according to the underlying surface conditions of the watershed. Finally， the Yuanzhou watershed in Yichun City， Jiangxi Province， is taken as a typical case for investigation， two kinds of rainfall losses， including vegetation interception and soil infiltration， are considered. The runoff coefficient is used to represent runoff loss process generated by rainfall. Then， the flood inundation data of Yuanzhou watershed at four different moments （10 minutes， 30 minutes， 1 hour and 2 hours） are evaluated by using the established cellular automata-based flood inundation analysis model. The results of the non-source inundation analysis method and MIKE SHE based-overland flow model are employed to validate the effectiveness of the established cellular automata-based flood inundation analysis model. The results indicate that the established model in this paper can well consider the influences of the underlying surface condition of the watershed and the rainfall production and confluence processes， and thus can simulate the flood inundation more reasonably and output the flood inundation data in real time， as compared with the non-source inundation analysis method and MIKE SHE based-overland flow model. The established cellular automata flood inundation analysis model has strong popularization and application prospects， clear practical physical meaning and simplified calculation process， and can accurately output the flood inundation process under a real-time rainfall event. The research outcomes can provide theoretical references for the risk assessment of flood disaster and the formulation of flood control and rescue decision plan under a heavy rainfall event.

Astronomical factor is one of the important factors in the regional river flow， but different astronomical factors have different degrees of influence and patterns on the runoff of the basin. To investigate the most important astronomical factors affecting the runoff of the Qiantang River Basin and their influence on the runoff of the basin， this paper identifies 26 astronomical factors such as sunspot relative number， lunar declination angle and 24 solar terms by using the methods of linear correlation， mutual information and gray correlation analysis and analyses the variation patterns between the major astronomical factor and the runoff of the basin. The main conclusions are as follows： ① Based on methods of linear correlation， mutual information and gray correlation analysis， the results of correlation analysis has shown that the lunar declination angle is the most influential astronomical factor on the inflow of the Qiantang River Basin； ②the runoff of the basin is abundant and the probability of wet year is 46.4% when the lunar declination angle changes from large to small， while the probability of dry year is 54.5%， and it is prone to extreme year when the lunar declination angle changes from large to small； ③the probability of dry year is about 66.7% and extreme dry year is frequent as well during the peak of lunar declination， whereas the extreme events happens frequently with 50% of the years being extreme during the trough of lunar declination. The results can provide a certain support for future long-term runoff evolution analysis and flood forecasting in the Qiantang River basin.

The construction and operation of reservoirs have changed the riverine hydrological regimes， causing adverse effect on riverine ecological health. Previous studies focus on ecological operation of reservoirs in large-basin plains， while few researches are about mountainous medium-small basins. To explore the relationships between ecological flow and water resources utilization of reservoirs in mountainous medium-small basins， this paper selects the two medium and a large reservoirs in the Majinxi Basin as a case study. A multi-objective ecological operation of reservoirs model is established to minimize amended annual proportional flow deviation （AAPFD） of the Wentu Station and total water deficit of the water region while maximizing the hydropower generation. The non-dominated sorting genetic algorithm （NSGA-Ⅱ） is applied to solve this model to research the relationship among the objectives of ecology， hydro-electricity and water supply under conditions of different inflow frequencies such as P=50%， 75% and 95%. The results show： ① with the increase in the inflow of the Majinxi Basin， the AAPFD of the Wentu Hydrological Station decreases， meanwhile the hydropower generation of reservoirs increases and the total water deficit of the water region declines， which improves the objectives of ecology， hydroelectric and water supply. ② There is a competitive relationship among the objectives of ecology， power generation and water supply and the ecology objective has a stronger competitive relationship with the power generation objective than the water supply objective； ③ the typical non-inferior solutions with the optimal objectives of ecology， hydroelectricity and water supply are selected as schemes 1~3， respectively， and there is large variability in schemes 1 and 2 of multi- objectives while the scheme 3 is similar to a compromise scheme compared with the schemes 1 and 2. The research results can provide a reference for the ecological operation decision-making of reservoirs in the Majinxi Basin.

Influenced by climate change， human activities， and other factors， the Haihe River Basin （HRB） is one of the regions with the most severe water resources decline in China. To study the influencing factors of the evolution of water resources in the HRB， taking into account the changes in precipitation intensity levels within the year， based on meteorological observation data， the third water resources assessment and water resources bulletin in the HRB， this paper uses the linear slope， M-K trend test， and t-test to analyze the variation in the total amount of water resources， different levels of precipitation， and potential evapotranspiration in the HRB from 1956 to 2019. The M-K mutation test is used to analyze the abrupt change time of the total water resources in the HRB， and the partial least squares regression is used to analyze the quantitative contributions of meteorological factors and underlying surface changes on the variations in the total water resources before and after the abrupt change. The results show that： ① The total amount of water resources in the HRB decreased significantly at a rate of 3.048 billion m³/10 a， and an abrupt change occurred in 1980. Before and after the abrupt change， the total water resources decreased from 41.09 billion m³ to 29.90 billion m³. ② Before and after the mutation， light rain， moderate rain， heavy rain， torrential rain， and potential evapotranspiration decreased by 8.34， 6.69， 12.96， 21.17， and 19.1 mm， respectively. Among them， light rain， moderate rain， heavy rain， and potential evapotranspiration decreased non-significantly， but torrential rain decreased significantly. ③ Among the 11.19 billion m³ of the decline in total water resources， the contributions of light rain， moderate rain， heavy rain， heavy rain， potential evapotranspiration， and underlying surface were -800 million m³ （7%）， -520 million m³ （5%）， -1.60 billion m³ （14%）， -3.47 billion m³ （30%）， 120 million m³ （1%） and -4.91 billion m³ （43%）， respectively. The reduction of torrential rainfall and the underlying surface changes were the main factors for the decrease in water resources in the HRB. The study serves as a scientific basis for formulating sustainable water resources development and utilization plan in the HRB.

There is a mutual feeding mechanism among water resources， economic society and ecological environment， which constitutes the water resources complex system. Exploring the coupling status of the composite system and analyzing the main influencing factors are of key significance to promoting the sustainable and coordinated development of the system in the future. Combined with the current characteristics and problems of Hunan Province’s water resources status， economic society development and ecological environmental protection， a coupled and coordinated evaluation index system of water resources， economic society and ecological environment system containing 28 indicators is established by using quantitative data analysis method， and the weight of indicators are determined by AHP-CRITIC combination weighting method based on game theory. The spatio-temporal evolution characteristics of the coupling coordination of the water resources， economic society and ecological environment system in 14 administrative regions of Hunan Province during 2005-2020 are evaluated through the coupling coordination model， and the obstacle degree analysis model is used to diagnose and analyze the influencing factors of coupling coordination degree of composite system. The results show that in recent years， the comprehensive evaluation index of water resources and ecological environment subsystems in Hunan Province has rebounded， and the overall comprehensive evaluation index of the composite system has shown an insignificant upward trend， and its change process is mainly affected by the evaluation index of water resources subsystem. At present， the coupling degree among the three subsystems is in a high level coupling state， and other regions except Xiangtan City are in a coordinated development state. The overall coupling and coordination degree of the system is basically stable at the intermediate coordination level， and generally develops in a good direction. According to the development and changing trend of the three subsystems and the identification of obstacle factors of the indicators， water resources and economic society have become the main factors affecting the future sustainable development of Hunan Province， which is the key control direction to solve the existing problems of Hunan Province and realize the future sustainable and coordinated development.

In recent years， global climate change and human activities have become increasingly severe. Studying the influencing factors of the dynamic changes of terrestrial water storage can help to deeply understand the temporal and spatial distribution of terrestrial water storage， and provide a basis for regional water resources management and protection. Based on （Gravity Recovery and Climate Experiment） GRACE data and eight climatic factors， this paper uses the trend analysis， cross-correlation analysis and （Partial least squares regression） PLSR model to analyze the （Terrestrial Water Storage Anomaly，TWSA） variation trends and the influencing factors in the South area of China （Yunnan， Guizhou， Sichuan， Guangxi， Hunan， Hubei and Chongqing） from the grid scale （1^o×1^o）.The results show that， from 2002 to 2017， the TWSA in most study area， except the western region， exhibited a significant growth trend. The highest increase rate reaches 0.9 cm/year in the junction area of four provinces （Chongqin， Guizhou， Hunan and Hubei） where the dominant pattern of TWSA variation in this region is guided by the long-term and seasonal trends. In Yunnan， central and southern Sichuan， southern Guangxi， and the eastern of Hunan and Hubei provinces， the dominant pattern of TWSA changes over time is seasonal variation. At the monthly scale， the TWSA dynamic is closest to the air specific humidity. 85% of the grid TWSA changes can be explained by climate fluctuations. The central and northern parts of the study area are significantly affected by the regulation of large reservoirs. In other areas， the influence of reservoirs on the annual-scale changes of TWSA in this area is not significant. At the annual scale， the change of TWSA in the whole region is still mainly affected by climatic factors. More than 60% of the change in terrestrial water storage can be explained by climate change， and rainfall is the main controlling factor for the annual change of TWSA in more than 70% of the study area. Only the vicinity of the Three Gorges Reservoir TWSA is still affected by a combination of anthropogenic activities and climate change.

Flow resistance is of great significance to the study of flow structure， bedload movement and transport rate. The factors affecting the flow resistance include bed sediment gradation， flow condition， bed surface structure and bedload transport rate. Considering a synthesis of bed surface sediment condition， flow condition and bedload transport， the flow resistance of flat and non-uniform bed surface is divided into grain resistance and bedload transport resistance. Based on the variation of the bedload transport rate on the process of bed surface coarsening， the degree coefficient of bed surface coarsening is expressed as a function of the relative transport rate. Combined with resistance formula of uniform bed surface and non-uniform bed surface which has coarsened completely， this paper establishes the grain resistance formula of bed surface including the structural parameters. Based on the difference of mechanical characteristics of rolling and saltation bedload， the forces of rolling bedload on the bed surface are analyzed， and the energy consumption of rolling bedload carried by water is obtained， and compared with the calculation formula which is established by bagnold. The bedload transport resistance formula is established by combining the proportion of rolling and saltation bedload under different flow strengths. The coarsening coefficient of bed surface is determined from the measured coarsening surface data， and finally the calculating formula for flow resistance of the flat and non-uniform bed surface with bedload transportation is established. Verified by the measured data and compared with other scholars’ resistance calculation formulae， the root mean square relative error （RMSRE） of the formula is only 0.094， which is less than that of other scholars， indicating that the resistance calculation formula is highly accurate.

The general polder is a unique polder form of Dongting Lake. However， with the development of social economy， the flood control protection objects of various general polders are quite different. It is urgent to further determine the classification of general polders in Dongting Lake during the flood control zoning， so as to formulate appropriate flood control measures.Using the geometric transformation of GCT graph， combined with population density， number of people protected by levees per kilometer， levee grade， flood storage loss， cultivated land area and importance parameters， Establish a model for judging the shape similarity of flood protection areas and flood storage and detention areas of general polders，230 general polders in Dongting Lake area were divided. The parameters of the 6-dimensional radar map of the reference value of the flood protection area are 900 people/km²， 0.46 people/km， Grade 2 embankment， 32 yuan/m³， 666.67 hectares of arable land， 2 points for importance. The parameters of the 6-dimensional radar map of the reference value of the flood storage and detention area are 600 people/km²， 0.18 people/km， Grade 4 embankment， 32 yuan/m³， 66.67 hectares of arable land， 1 point for importance. Use Matlab to normalize the data， divide the 6 parameters of each general polders into dimensionless ones within the range of 1~5， and draw a 6-dimensional radar map. At the same time， the similarity judgment of the 6-dimensional radar map with the reference value of the flood protection area is completed， and the judgment result is general polder that is not similar， continuing to determine the similarity with the 6-dimensional radar map of the reference value of the flood storage and detention area. If it is still not similar， it will be determined as a flood area. The results show that there are 67 flood protection areas， 53 flood storage and detention areas and 110 flood areas. Through a further analysis of the results， suggestions on model modification and subsequent general polders planning are put forward. It serves as a reference for the division of flood protection area， flood storage and detention area and flood plain to be carried out in the general polders of Dongting Lake Area.

Vegetation coverage is an important indicator of land ecology. The Yellow River Basin is an important national ecological barrier. It is also an important economic zone and energy basin. In order to reveal the changes in the ecological environment of the Yellow River Basin， based on the GIMMS NDVI data， land use data， topographic data and climate data and Mann Kendall， Sen method and geographical detectors， this paper explores the spatio-temporal dynamics of seasonal NDVI in the Yellow River Basin in recent 40 years and the contribution of different factors to seasonal NDVI changes. The results show that ① the NDVI value in arid areas was the lowest， the second was the semi-arid area， the humid area had the highest NDVI and the second was the semi- humid area. ② The increase rate of NDVI in spring was 0.001 2/a. The NDVI change in 2005 was abrupt. The increase rate of NDVI in summer was 0.001 4/a. 2008 was the year of NDVI mutation. The increase rate of NDVI in autumn was 0.001 5 /a. The year 2002 was the mutation year of NDVI in autumn. The increase rate of NDVI in winter was 0.000 4 /a. 2007 was the year of NDVI mutation. ③ The increase rate of NDVI decreased with the increase in altitude， especially in summer. The slope had little effect on the change rate of NDVI in spring and winter， but the change rate of NDVI in summer and autumn decreased significantly with the increase in slope. ④The q influence of different factors on NDVI was the q determinant of the minimum value of daily maximum temperature （TXN） > the maximum value of daily minimum temperature （TNX） > the minimum value of daily minimum temperature （TNN） > average annual temperature （TEM） > land use （LUCC） > the maximum value of daily minimum temperature （TXX） > the maximum value of rainfall for five consecutive days each year （RX5DAY） > wind speed （WIN） > altitude （DEM） >SLOPE> average annual cumulative precipitation （PER）. The interaction of any two factors was more than or equal to the effect of a single factor on NDVI. The q determinants of TEM and per on NDVI decreased with time， while the q determinants of LUCC and SLOPE increased. Extreme temperature and precipitation have no obvious time trend.

The study of maintaining the critical channel slope of the middle-water channel in the tail reach of the Yellow River Estuary can provide effective parameters for the estuarine channel regulation project， and is of great significance for maintaining the stability of the flow path of the Yellow River Estuary. This paper explores the critical channel slope of the middle water channel in the tail reach of the Yellow River Estuary. Firstly， on the basis of the research results on the principle of the minimum work of the river bed evolution， the mathematical expression of the minimum work average velocity of the reach is derived， and the functional relationship between the minimum work average velocity of the reach and the inflow and sediment conditions， the flat discharge of the tail reach and the channel slope in the quasi-equilibrium period is calculated and analyzed. Finally， the channel slope range which should be controlled to maintain the flat stream discharge in the middle water channel of 4 000 m³/s in the sink section of the Yellow River estuary is estimated， according to the established functional relationship.The results show that， the correlation coefficients between the minimum average velocity of the tail reach of the Yellow River Estuary and the average flood season discharge， sediment concentration， sediment transfer coefficient and the plain discharge in the same period are all more than 0.9. According to the calculation results， the critical channel slope range of the middle water channel in the tail reach should be between 0~0.087.

River habitat connectivity is very important for the quality and stability of river water ecosystem， and is the key to river ecological recovery and health protection. How to quantify the impact of the dam site on river habitat connectivity is of great significance for water ecological protection and restoration. In this study， North Canal River， a typical multi-dam river in northern China， is selected as the research object， and the influence of dams on river habitat connectivity is quantitatively analyzed by Dendritic Connectivity Index （DCI）.The relationship between the number of dam space， the length of dam cut section， the length of total river length and DCI value in the basin in 1980， 2000 and 2020 shows that the number of dams in different periods was 11， 20 and 23， with DCI values of 51.6， 28.7 and 26.5， respectively. During the 40 years from 1980 to 2020， the number of dams increased by 109.1%， while DCI decreased by 48.6%， reflecting the deterioration of river habitat connectivity. The results also show that the impacts of dams in different spatial locations on river habitat connectivity are different， which shows that the dams in the main stream have more impact on river habitat connectivity than those in tributaries. The simulation results show that the average impact of dams in the main stream of the North Canal River is about 10 times that of the tributaries. The fish species and quantity survey also show that the increase in the number of dams in the North Canal River has a significant negative impact on the river habitat connectivity. In this paper， the analysis of habitat connectivity guarantee of river basin by increasing the DCI value of dam has certain guiding significance for ecological recovery， protection and restoration of multi-dam rivers.

Field and laboratory experiments are usually used to reveal the impacts of the slope shape on the soil erosion process， but the fixed experimental conditions are restrained in analyzing the impacts of other related factors on the slope shape effect. Based on the widely used WEPP model， the soil erosion processes are simulated and compared for four common slope shapes in nature， including linear， convex， concave， and S-shaped. The sensitivities of slope shape effect to different rainfall intensity， soil erodibility， and rill density parameters are also analyzed. The results show that soil erosion and its spatial distribution can be significantly changed for different shaped slopes. With the increase in rainfall intensity， both the total erosion and its spatial difference tend to increase greatly for four shaped slopes. The increase in total erosion for the convex slope is more significant， which is about two times as large as that of the concave slope. As the ratio of rill spacing to initial rill width increases， the soil erosion intensity and its spatial difference decrease on all slopes. The decrease in soil erosion intensity is more significant on S-shaped slope but less significant on linear slope. With the increase in rill erodibility parameter， the soil erosion density and its spatial difference increase on four slopes. The spatial difference on convex and S-shaped slopes changes greatly in comparison with those on linear and concave slopes. However， the soil erosion density and its spatial difference are not sensitive to the interrill erodibility parameter for all slopes. The above results are helpful to deepening the theory of soil erosion on different shaped sloping surfaces， which are expected to be useful for soil erosion prevention and high-quality development of sloping farmland on Loess Plateau.

The accurate remote sensing classification of surface coverage in reservoir areas provides important basic data for reservoir ecological environmental protection， reservoir optimization and dispatching， and soil and water conservation. In recent years， deep learning has been widely used in surface coverage classification. The current surface coverage classification method based on convolutional neural networks is usually a local spatial operation using regular convolutional kernels， which cannot easily capture long-distance dependencies between cells. Therefore， this paper proposes a dual Graph Convolutional Fusion Network （Dual-GCFN） for remote sensing classification of surface coverage. The network is divided into two parts. The upper branch uses the K-neighbor algorithm to construct a topological diagram between cells， and then uses the Graph Neural Network （GNN） to aggregate neighborhood features to update the target node information to achieve the aggregation of spectral features； the lower branch uses a dense Convolutional Network （DenseNet） to convolve local blocks of images to extract spatial information and high-order semantic information of remote sensing images. Finally， the spectral characteristics and spatial characteristics will be effectively fused through the dimensional channel Add operation， and after the fully connected layer， the Softmax classifier is used to realize the surface coverage classification. The Chushandian Reservoir area in Henan Province is the research object， and it uses Sentinel-2 remote sensing image data and DEM data in the research area to conduct experiments. The results show that the overall accuracy and Kappa coefficient of the Dual-GCFN classification proposed in this paper are 90.63% and 0.87， respectively. Compared with the classical classification method Random Forest classification （RF）， single GNN or DenseNet classification results， the overall accuracy increases by 15.97 percentage points， 10.86 percentage points， and 4.65 percentage points， respectively， the Kappa coefficient increases by 0.26， 0.24， and 0.08， respectively， and the misclassification of ground objects and the point noise in the classification map are greatly reduced in the classification map.

Construction standard of water and soil conservation project in the mountainous area of South Ningxia is low， the aging of the project is serious， and the utilization efficiency of local water resources is low. In order to solve this problem， a closure and slow-release ecological system has been built locally. The closure and slow-release ecological system is composed of a series of closure and slow-release ecological dams arranged reasonably in space. The closure and slow-release ecological dam is a water and soil conservation measure for improving small silt dams. Its main biological and engineering measures have synergistic effects on watershed hydrology. In order to reveal the impact of closure and slow-release ecological system on the runoff of small watersheds and provide a reference for the new development of the silt dam in the future， CASC2D-SED model of the watershed is constructed by taking camel forest watershed as the research object， the parameters are calibrated and verified based on measured rainfall-runoff data， the changes of underlying surface conditions after the project is completed are determined by field investigation and test， and these conditions are substituted into the model to couple the project with the model. By comparing the measured and simulated comparison and analysis of runoff before and after the project construction， and combining the calculation and analysis of runoff with runoff coefficient of the river basin， the influence of the project on the river basin runoff is determined. After the model is built， the deterministic coefficient of the rate periodicity and verification period are above 0.7， the absolute value of relative error of flood peak is within 20%， and the fitting degree is good. According to the calculation and analysis of runoff and runoff coefficient of the river basin， after the construction of the project， the runoff and runoff coefficient of the river basin have increased significantly， and the runoff coefficient has increased by about 3.5%. The basin model is used to simulate the basin flood process before project construction， and the coupling model is used to simulate the basin flood process after project construction. The results show that the project can effectively reduce the peak flow and reduce the peak flow by about 30%. The closure and slow-release ecological project can significantly increase the runoff coefficient of the watershed and reduce the flood peak. So， the closure and slow release ecological system can increase the water utilization efficiency and tap the potential of water resources. Moreover， due to the convenient construction and low cost of the project， it can be popularized to a certain extent.

Pond wetlands are widely distributed in the southern irrigation areas in China， and a reasonable wetland design is beneficial to improving their comprehensive benefits. This paper analyzes the effects of plant species （Schoenoplectustabernaemontani， Thalia dealbata， Pontederia cordata） and inlet flow rate （4 m³/h and 11 m³/h） on the treatment performance and hydraulic performance of pond wetlands through static and dynamic removal tests and tracer tests in the Erhai watershed of Dali City as a typical area. The results show that： ① The nitrogen and phosphorus treatment performance of Pontederia cordata wetland is better than that of Schoenoplectustabernaemontani wetland， and better than that of Thalia dealbata wetland； the hydraulic performance of Schoenoplectustabernaemontani wetland is better than that of Pontederia cordata wetland， and better than that of Thalia dealbata wetland； the combined treatment performance and hydraulic performance of Schoenoplectustabernaemontani is better than that of Pontederia cordata， and better than that of Thalia dealbata. ② When the inlet flow rate is reduced from 11 m³/h to 4 m³/h， the total nitrogen removal rate of the wetland increases from 17.8% to 32.7%， the total phosphorus removal rate increases from 19.9% to 28.6%， the effective volume ratio increases from 0.60 to 0.77， the Morrill dispersion index increases from 3.12 to 3.59. It indicates that the treatment performance and hydraulic performance of the wetland are improved， and the short-circuiting of water flow is reduced， but the degree of mixing has been increased. ③ The pollutant removal rate of the first half of the wetland accounts for 65% to 87% of the overall， and the effective volume rate is above 0.85， with better treatment performance and hydraulic performance than the wetland as a whole.

Flow velocity is of great significance to understanding the process of inter-rill erosion and hydrology on a hill slope under rainfall conditions， which is an important parameter for hydrology and soil erosion modeling. A solute centroid method is employed to measure the velocity distribution of sheet flow based on solute transport process， to understand the influence of straw mulching on flow velocity. The measurements are made at slope lengths of 3， 5， 6 and 7 m under rainfall intensities of 80， 120 and 160 mm/h， straw mulching quantities of 2 and 8 t/hm²， and slope gradient of 10°. The measurements clearly indicate that flow velocity is significantly affected by rainfall intensity and straw mulching quantity. In this paper， the velocity of sheet flow ranges from 0.016 to 0.108 m/s， increases with rainfall intensity， and decreases with straw mulching quantity. Flow velocity increases with the slope length， while the increase rate slows down before approaching a steady velocity. The distance for sheet flow velocity to reach its maximum becomes shorter under greater rainfall intensity and straw mulching quantity. This paper helps understand the hillslope hydrodynamics along the mulched slope under rainfall conditions.

In large-scale inter-basin open channel water transfer projects， all kinds of water conveyance elements are connected with each other， and their flow characteristics are different， an analysis of the hydraulic response characteristics of the ditch basin is of great significance for the safe and stable operation of the project. By constructing one-dimensional hydrodynamic model and taking safety regulation as guidance， the feasible time domain of safety regulation is analyzed. In order to avoid the occurrence of water level exceeding the limit， the earliest and the latest regulating time of the buildings that do not exceed the upper and lower limits of the water level of the canal pool are calculated， taking the upper and lower reaches of the pumping station in the river mouth of the Yangtze-Huaihe River Diversion Project as an example， the study is carried out. During the study， due to the selection of channel segment， the earliest regulation time was 0 h， so the latest regulation time is analyzed. The results show that： ① when the initial upstream flow and downstream water level remain unchanged， the latest regulation time decreases with the increase in the target flow and flow variation in a power function relationship ② When the initial upstream flow is the same and the flow variation is different， the latest regulation time gradually decreases with the increase in the downstream water level， and the decrease trend of the latest regulation time becomes gentler with the increase in flow variation； ③ The latest regulation time has the strongest correlation with the flow variation， but has a poor correlation with the upstream initial flow. Finally， based on the calculation results of feasible region of safety regulation time， a fast generation formula of the latest regulation time is constructed.

In practice， overemphasizing construction but disregarding management is particularly highlighted in rural small water conservancy facilities. The traditional mode of collective maintenance is disjointed from the current family-contract responsibility system. In some areas， the phenomenon of 'the government's management is not in place， the collective cannot manage， and the peasants are unwilling to manage' even appears in small rural water conservancy projects， which seriously affects the normal operation of the utility of these projects and becomes the bottleneck for the development of agricultural modernization and the promotion of farmers′ income increase. Therefore， the post-management of small rural water conservancy projects has attracted much attention. Based on the current situation of the management and protection of small-scale water conservancy projects in rural areas， this paper takes a small-scale water conservancy project in a county in southeastern Chongqing as an empirical analysis object. The central theory of the post-management mode of small rural water conservancy projects in Chongqing is systematically discussed， and policy recommendations for reference are put forward to improve the efficiency of the post-management of small rural water conservancy projects.

The production and transportation of abscisic acid ABA in potato roots and leaves under different drip irrigation conditions were explored， as well as the regulation of stomata and transpiration by ABA， which provided a basis for physiological water saving under drip irrigation conditions. Five gradient irrigation treatments of 3 000~6 000 m³/hm² were set up to determine the microphysiology of potato leaves and the content of abscisic acid in root leaves under different treatments. The results showed that potato stomatal conductance， transpiration intensity， and net photosynthetic rate were positively correlated with irrigation volume， and showed a significant linear relationship with irrigation volume， while the intercellular CO₂ concentration was the opposite. Leaf abscisic acid ABA plays a certain role in regulating leaf stomatal behavior， but its determination coefficient is only 0.72， indicating that there are other factors or endogenous hormones that can regulate leaf stomatal behavior together with abscisic acid ABA. It shows that leaf abscisic acid ABA plays a direct role in inhibiting potato transpiration. When soil moisture is insufficient， potato produces endogenous hormone ABA through the body's adaptive feedback system， which reduces leaf transpiration intensity and reduces water loss.

For a comprehensive analysis of magnetizing the influence of water irrigation on grapes under salt stress， the experiment was carried out from April to October 2021 in Ningxia central test base. The two factors of irrigation water salinity and magnetization treatment were set. The irrigation water salinity was set to four levels （S1~S4）， and the magnetization treatment was set to two levels （M， N）， and，the conventional water source was the control （CK）. So， a total of 10 treatments were used to analyze the response characteristics of soil salinity accumulation in the root zone of grapes， physiological growth of grapes， and yield and quality. The results showed that compared to the non-magnetization treatment， the magnetization treatment could reduce the soil water storage in the 0~100 cm soil layer by 1.5~21.9 mm， and reduce the salinity accumulation in the 0~20 cm soil layer by 13.04%~29.63%. However， when the mineralization was too high， the magnetization treatment did not significantly improve soil moisture and salinity. Salinity stress would affect the branch growth， stem diameter and leaf area of grapes. The leaf area of M_CK， MS₁ and MS₂ was higher than that of N_CK， NS₁ and NS₂ treatments improved by 1.06%， 4.55% and 27.51%， respectively， but when the mineralization was low， magnetization could improve the physiological characteristics of grapes， promote photosynthesis， transpiration rate and stomatal conductivity， but at higher salinity stress； mild salinity stress can promote the increase in the grape output and WUE， but with the stress deepening； magnetized water irrigation can significantly increase the grape yield and WUE. Magnetization promotes grape quality under salt stress， but the effect was not significant. After a principal component analysis， the comprehensive score of the treatment on grapes ranges from high to high： MS₁>MS₂> MS_CK>NS_CK>MS₃>NS₁>MS₄>NS₃>NS₂>NS₄.

Combined with an inverted siphon pipe in the Water Diversion Project in Central Yunnan， the research on high strength steel Q460C to be used in the industries of water resources and hydropower is carried out through a 1∶1 large scale model test. The stress state and ultimate bearing capacity of the large diameter water pipe， as well as the changing law of stress and displacement during the water pressure varying from zero to the end of the blasting are obtained. Model test results show that before the hydrostatic test pressure 2.82 MPa， steel pipe stress/strain~pressure and displacement pressure curve show a good linear relationship， and the maximum stress is less than Q460C steel allow stress under the working condition of hydraulic pressure test， and the penstock works in an elastic state and can meet the requirements of strength design. The maximum blasting pressure of the model is 7.5 MPa， which is 3.3 times of the design pressure of 2.26 MPa， and a large plastic deformation is produced. It shows that the penstock made of Q460C steel has good toughness and enough safety margin， and can be applied to the inverted siphon steel pipe structure of the water diversion project in central Yunnan or similar projects.

It is easy to form a contradiction between the drainage of pumping stations based on people’s livelihood and water intake of waterworks. In particular， the actual problems such as whether the drainage station can be built within the quasi-protection zones of the water intake or how far the distance between the two needs to be scientifically demonstrated. Physical model is used to study the influence of the pumping station drainage on the downstream water plants in the tidal reach of the Yangtze River. A qualitative simulation experiment using potassium permanganate tracer and the semi-quantitative simulation experiment measuring conductivity with NaCl solution is carried out. The results show that potassium permanganate qualitative tracing could observe the flow trace of drainage water at different Yangtze River flows， with the increase in the Yangtze River flow， the flow trajectory discharged by the pumping station deflected regularly from the deep trough to the onshore direction due to the current jacking. The semi-quantitative simulation experiment data show that the influence of drainage water body on the conductivity of water intake shows an obvious regular pattern of two rises and two falls. And because of the superposition effect， the latter peak is higher than the previous peak. When the drainage flow of the pump station is 21 m³/s， the maximum dilution ratio of the conductivity of the water intake is 1.58%， 2.46% and 2.12% respectively under the three stages of Yangtze River flow from large to small. When the drainage flow of the pumping station is 14 m³/s， the maximum dilution ratio of the conductivity of the water intake is 0.92%， 1.31% and 0.94% respectively under the three stages of Yangtze River flow from large to small. Only the drainage flow of the pumping station is 7 m³/s， the drainage flow had no impact on the water quality of the water intake. The pumping station drainage had the greatest impact on the water intake when the middle flow of the Yangtze River and the pumping station drainage flow are the largest. It can be seen that the pumping station drainage has the greatest impact on the water intake when the balance between the Yangtze River flow scouring dilution and the water trace direction is realized. In addition， the physical model can be applied to the study of pollutant diffusion effects in the environmental field.

At long-distance and high-lift pumping stations， serious phenomena of the water hammer such as water column separation and transient pressure along pipelines of water conveyance systems when pumps stop due to failures. In view of this， based on the theory of fluid transients and the method of characteristics， with a mathematical model of water column separation， this paper deduces boundary conditions for axial flow check valves and builds a mathematical model of the water conveyance system. For the application of axial flow check valves in engineering projects， measures and effects of protection are modeled numerically. Moreover， the working conditions at different installation positions of the axial flow check valve are numerically simulated. After that， the comparison of changes in the transient pressure of the pipeline shows that the axial flow check valve shall be set in the middle and rear sections of the pipeline at the long-distance high-lift pumping station. What’s more， it is also concluded that the position of closing the high-pressure water hammer can be produced after pump stopping. Hence， applying the axial flow check valve as the water hammer protection of the pumping station has a new idea for the long-distance high-lift water conveyance system.

Through the numerical simulation of the shaft inlet channel of a pump station， it is found that the flow separation phenomenon occurs at the confluence of the shaft tail. From the perspective of the line shape and size of the shaft tapering section， 15 optimization schemes are designed to study the hydraulic performance of the shaft inlet channel and the flow separation phenomenon at the confluence of the shaft tail. The optimal scheme is studied by the model test. The results show that the flow separation at the tail confluence of the shaft inlet channel will reduce the weighted average angle of the outlet section velocity of the channel. Under the same linear scheme of the shaft tapering section， the hydraulic loss of the shaft inlet channel first decreases and then increases with the increase in tthe shaft tapering section length， and the velocity weighted average angle gradually increases with the increase in the shaft tapering section length. When the linear optimization scheme 2 is adopted， there will be no flow separation at the confluence of inlet channel and shaft tail. Among the 15 schemes， the hydraulic performance of the inlet channel of scheme JS14 is better， which is the preferred inlet channel scheme for a pump station. The maximum efficiency of model test of pump device adopting scheme JS14 can reach 77.57%， and the numerical simulation of the hydraulic loss in the inlet channel of the device is consistent with the model test.

To investigate the transient working characteristics of the pump， the opening valve law and the pressure change in the pipeline in the filling process of the pumping station， the rigid column model and elastic column model of the hydraulic transient of the filling process are established based on the wave tracking method and the interface tracking method. The calculated and experimental results of the filling process of a pumping station show that the hydraulic transient of the filling process in the short pipe calculated by the two models is consistent with the experimental results. In the process of valve opening， the flow rate of the pump increases sharply， the head decreases rapidly and the power increases. After reaching the extreme value， with the completion of the filling process， the pump parameters are stable at the working point. The pressure head after the valve is the result of a combined effect of primary and reflected waves.

To explore the influence of sediment on the Basher trough under different channels and slopes. In this paper， the method of combining numerical simulation and experiment is adopted to lay natural river sand in the upstream of three channels under different slopes， stir them evenly in the water tank， keep their sediment content unchanged， and then set different inflow flows. The changes of sediment distribution， velocity distribution， flow measurement error and sediment carrying capacity in Basher trough under different schemes are analyzed. At the same time， the “fluent” software is used to carry out three-dimensional three-phase flow numerical simulation to verify the physical experiment. The results show that the water and sediment simulation by Euler model is consistent with the results， and the maximum error does not exceed 4.23%. In the Basher trough test， the maximum flow measurement error in all working conditions is 15.08% at the gradient of 0.8% in the U-shaped channel. The minimum flow measurement error is 2.13% at the gradient of 0.1% of the trapezoidal channel. Among the three channels， the flow measurement error under trapezoidal channel is the smallest， and the flow measurement error under U-shaped channel is the largest. The Basher trough of trapezoidal channel has the highest flow measurement accuracy， and the sediment distribution is concentrated on the upstream connection section of the Basher trough and both sides of the channel bottom. The trough has less sediment， which has little influence on the flow. The flow measurement accuracy can be satisfied well in the Basher trough with a slope of 0.1% to 0.4%， while the flow measurement error is larger in the Basher trough with a slope of 0.4% to 0.8%. At the same time， when the maximum measured water level is 34.44% to 53.33%， the flow measurement error of the Basher trough decreases rapidly. When the flow measurement error exceeds 53.33%， it decreases slowly， and finally the curve tends to be stable， and the flow measurement error is relatively small. When the water level of Basher trough is less than 15.49% of the maximum measured water level， the flow measurement error is relatively large. In the case of large flow， the sediment carrying force in most parts of the Basher trough is more than the average sediment content of the channel， indicating that there is no siltation phenomenon in the Basher trough in the case of large flow. In the case of large flow， the diffusion section has the maximum sediment carrying force， while in the case of small flow， the throat section has the maximum sediment carrying force.

In order to solve the problem of regional water shortage and water resource utilization， China has planned to build a complex water diversion system with multiple pumping stations in recent years to realize trans regional water transfer. This kind of project has the characteristics of long distance， multiple pumping stations， multiple water supply terminals， complex pipeline system， etc. It is difficult to achieve the goal of optimal dispatching and economic operation. Through a research on the key technologies of intelligent dispatching of the complex water diversion system in the Qingshui River basin of Ningxia， under the premise of ensuring the safety and quality of water supply， this paper adopts the method of combining mathematical simulation with on-site monitoring， through an intelligent analysis and optimal dispatching calculation， to quickly obtain an optimal operation plan， to achieve intelligent analysis， intelligent diagnosis， economic operation and dispatching decision-making of all stations along the line， and to improve the overall operation economy and operation efficiency of the system， effectively improve the assurance rate of water supply to achieve safe， efficient and economic operation.

The existence of cavitation is very unfavorable to the operation of the turbine， which will not only reduce the output and efficiency of the turbine， destroy the overcurrent parts of the turbine， but also lead to the unsafe and unstable operation in serious cases. Accurate identification of the cavitation position helps to prevent cavitation damage. This paper uses the VOF multiphase flow model and SST k- \omegaturbulence model to simulate the Schnerr-Sauer cavitation model. Through an analysis of the distribution rules of the velocity field and pressure field of the gas phase are obtained. The main conclusions are as follows： When the fluid flows through the contraction pipe section， the speed increases rapidly， and the absolute pressure at the outlet of the nozzle drops rapidly to the saturated steam pressure needed to form cavitation， which creates conditions for cavitation. There is a speed attenuation in the center of the injection needle， and a low pressure zone in the nozzle exit and the tip area of the injection needle. The gas phase is mainly distributed at the outlet position of the nozzle. Under the opening of 20% and 80%， the gas phase is symmetrically distributed in the injection outlet torus， but not fully covered， the gasification area corresponds to the position of the anterior fluid diversion area， and the opening of 40% and 60% evenly covers the nozzle outlet. The maximum gas phase generation rate and gas distribution along the axial direction under different injection needle opening degrees first increase and then decrease with the opening degrees. The water vapor content at 40% opening degree is the most， and the cavitation range at 40% opening degree is the largest

The traditional hydro-generator unit maintenance training method has problems such as low efficiency， long period， and certain dangers in field teaching， and its training effect is difficult to guarantee. Therefore， a virtual reality-based three-dimensional visual training method and training platform for the maintenance， disassembly， and assembly of hydro-generator units are proposed. This method takes the Francis turbine-generator unit of the Three Gorges Hydropower Station as the simulation object， uses 3dsMax to create the component-level three-dimensional maintenance model， and constructs the interactive scene of three-dimensional visual maintenance， disassembly， and assembly based on Unity3D. It designs and implements the functions of a three-dimensional visual display of maintenance process knowledge， the interactive practical operation of maintenance， disassembly and assembly， and online scoring of maintenance skill assessment. It solves the problem of accuracy evaluation of uncertain maintenance steps through the AHP-fuzzy comprehensive evaluation method and physical space constraints of maintenance components and constructs a complete closed-loop of independent learning for the maintenance training of hydro-generator units. The practical application shows that the platform satisfies the requirements of maintenance training of hydro-generator units and provides effective technical support for large-scale standardized maintenance of hydro-generator units.

In order to ensure the water， power matching relations between upstream and downstream of cascade hydropower stations， under power market conditions， it is necessary to carry out researches on short-term optimization operation mode of cascade power stations facing the spot market and exploit the matching method of water quantity and electricity quantity under the unified bidding mode of cascade power stations. This paper focuses on electricity spot market oriented joint optimization scheduling problem. Based on an analysis of the correlation between the power demand and the clearing price in the power market， the paper proposes using the power demand process to drive the optimal operation of cascade hydropower stations. By taking the cascade hydropower stations from Pubugou reservoir to Tongjiezi reservoir on Dadu River as an example， this paper constructs an expected power maximization model considering price-fluctuation within a day. A newly proposed dragonfly algorithm （DA） is applied to model calculation. A typical day data is selected as case study. The simulation results show that ① the operation modes such as water level process and output process of the cascade power station obtained by DA conform to the general dispatching laws， the calculation time and efficiency meet the practical requirements （about 120 s）， these indicate that DA can be used to solve the short-term optimization scheduling problem of cascade hydropower station with high dimension； and ② there is a strong correlation between power demand and clearing price， especially the occurrence time of peak and trough of the two is highly consistent. It indicates that the optimal operation of cascade hydropower station driven by power demand can guide the cascade hydropower station to generate more power during the period of large demand. Therefore， the expected power generation maximization model established in this paper is feasible， which can achieve the pursuit of greater economic benefits while ensuring the water， power matching relations between upstream and downstream of cascade hydropower stations. It can be used as a reference for the electric quantity declaration scheme of the hydropower system participating in the spot market bidding.

The characteristic method is used to calculate the large fluctuation transient process of a diversion power generation system with upstream throttled surge chamber. The influence of the length of headrace tunnel， closing law of guide vanes， roughness of pipeline and moment of inertia of the unit on the value of ideal throttled orifice area in the hydraulic transients are simulated and analyzed. The results show that the larger the length of the headrace tunnel， the moment of inertia of unit and the roughness of pipeline， the smaller the area of ideal throttled orifice. The conclusion has some guiding significance for practical engineering design.

In order to ensure the safe and stable operation of the hydropower plant units and optimize the maintenance strategy， this paper proposes a prediction algorithm for the remaining life of the head cover drainage pump for the water turbine head cover drainage system. On the basis of real-time data， combined with the reliability principle， the corresponding operation reliability of the head cover drainage pump is calculated according to its vibration speed， vibration displacement， pumping flow and pumping time， and the integrated operation reliability of the head cover drainage pump is obtained. Combined with the limit learning machine algorithm， the residual life assessment model is constructed， and the dynamic residual life of the top cover drainage pump of the hydraulic turbine is comprehensively predicted. Combined with the real-time data of the top cover drainage system of the hydropower plant， the effectiveness and feasibility of this method are verified.

In recent years， the amount of hydropower development is increasing day by day. In order to save investment， many small and medium-sized hydropower stations with long diversion channels choose to use pressure regulating valves instead of surge shafts to cope with the large fluctuation transition process. However， setting pressure regulating valves cannot reduce the inertia time constant of water flow of units， which will lead to poor dynamic characteristics in the small fluctuation transition process. In order to find a way to improve the dynamic characteristics of the turbine regulating system of the “valve instead of well” hydropower station in the small fluctuation transition process， the small fluctuation disturbance simulation of a conventional hydropower station with a long pressurized diversion channel is carried out， and the dynamic response of the regulating system in the small fluctuation transition process is analyzed. The dynamic response results of the regulation system show that the poor dynamic characteristics of the turbine regulation system in the conventional hydropower station with a long pressurized diversion channel are due to the excessive response speed and amplitude of the servomotor caused by the excessive speed of the governor. A new hydraulic turbine control strategy is put forward， which feeds back the displacement signal of the servomotor to the proportional and differential links of the governor. By restraining the speed and amplitude of the servomotor action， the dynamic performance of the “valve instead of well” hydropower station in the small fluctuation transition process can be improved. Through simulation comparison， this method can effectively improve the dynamic performance of the small fluctuation transient process of the turbine regulating system of “valve instead of well” hydropower station.

In construction projects， EPC is a method widely used at home and abroad， but it is not widely used in the construction of water conservancy and hydropower megaprojects in China. Based on the analysis of the construction characteristics of water conservancy and hydropower megaprojects and the development of China's hydropower construction market， the main issues of EPC project transaction governance structure are optimized， including the use of the “total price” pricing method for the project as a whole， but for some sub-projects with large uncertainty of “site data”， the pricing method with risk allocation is adopted to motivate the contractor to optimize the project. And the general contractor of the project is required to build a tight consortium， namely cooperative enterprise， so that the project design and construction can be deeply integrated to fully exploit the potential of market entities and give full play to the advantages of the general contracting transaction method. The effectiveness of the above optimization methods are demonstrated by two work cases.

In view of the problem of determining the parameters in the formula of PanJiazheng to calculate landslide surge in the reservoir area， the sensibility analysis of the number of series items and the wall reflection coefficient to the calculation of surge height is discussed. In this paper， Pan Jiazheng method is used to calculate the surge process of Bianjiazhai landslide in the reservoir area of Suofengying Hydropower Station. The results show that the calculation of surge propagation and height is sensitive to the selection of the number of series items， which increases with the increase in the number of series items， and the difference is large. When values are selected according to the table， the characteristics of landslide movement and water surface in the reservoir area should also be considered. Calculation is not sensitive to the wall reflection coefficient， and the error between the proposed value range and the median result is within 0.6%. When the wall is not clear， the median value can be taken.

Late-stage support plays an important role in improving the per capita disposable income of reservoir immigrants. Therefore， it is of practical significance to quantitatively analyze the present situation of income increase supported by reservoir resettlement in the later period and formulate reasonable planning suggestions. In order to reveal the lag effect of input indicators on the change of migrant income and reflect the actual performance level of late support， this paper calculates the lag period of input indicators by time difference correlation analysis method， and establish a DEA model of late support lag effect. From three aspects of resettlement quality， production conditions and capital investment， three sample counties in YF Citing Province are taken as research objects to study the support performance and lag effect of reservoir resettlement in the later stage. The results show that the post-migration support in YF city is not DEA effective， with the average comprehensive efficiency of 0.931 and the scale efficiency lower than the pure technical efficiency. It has the characteristics of low technical management level and excessive investment scale in the county， and the average efficiency will decrease by 0.5%~14.4% without considering the lag effect. The results of the study objectively reflect the situation of support in the later stage， and can provide a basis for the implementation and adjustment of the support policies in the later stage of migration.

Aiming at the problems of poor visualization effect and single functional interaction mode of the digital management system for hydraulic metal structure safety monitoring， this paper designs a virtual system for hydraulic metal structure safety monitoring based on virtual reality technology. The system through the remote data access， the real-time acquisition of radial gate， the hydraulic hoister such as sensor data， through the RabbitMQ message queue associated with the virtual scene， complete the data driven by real-time simulation， and monitoring the chart data visualization， which realizes the spillway hydraulic metal structure safety monitoring and the integration of virtual reality technology. And the system is applied to the actual hydraulic metal structure safety monitoring through virtual roaming， model interaction and dynamic simulation. The application shows that the system has advantages like strong immersion and good interactivity， and can monitor the operation status of gates and hydraulic openers in real time， which helps managers to make accurate judgments on the safety operation and maintenance of hydraulic metal structures at the later stage. It serves as a valuable reference for the intelligence management of hydraulic metal structures.

The analysis of thermal-humid environment distribution of underground hydropower station is of great significance for plant operation. This paper， by using CFD considering outdoor weather， indoor heat and wet source， studies the distribution law of thermal-humid environment and through online and offline monitoring results of generator layer which shows simulation precision. The results of simulation is used to investigate the distribution of thermal-humid environment of the generator layer as well as the analysis of thermal comfort parameters of air distribution. Results show that the three kinds of meteorological parameters of thermal-humid environment under the condition of similar distribution， uniform temperature field distribution results in uneven temperature coefficient value， velocity field and humidity field changes obviously， so the speed of non-uniform coefficient is large， three weather conditions of energy utilization coefficient are large that reflects better energy saving potential.