To accurately obtain the health performance level of a pumped storage unit (PSU), a health performance tendency prediction method based on convolution neural network-long short-term memory neural network (CNN-LSTM) is proposed. Firstly, a unit health state model based on Gaussian process regression was constructed to effectively characterize the operating characteristics of the PSU. Then, an index that can quantify the health performance of the PSU was proposed. Finally, by integrating the good local feature extraction ability of the CNN and the advantage of the LSTM in time series prediction, a prediction model based on CNN-LSTM was proposed. The experiments were conducted using monitoring data from a pumped storage station in China. The results show that the proposed method can betterly predict the future evolution of the PSU's health performance.

Regarding the impact of the construction of the Guotan Hub of the Tanghe Navigation Project on the flood control safety of the river channel, the design flood of Guotan Hub was calculated through measured flood data of hydrological stations and linear interpolation method. According to the measured cross sections and thalwegs over the years, the river channel scouring and silting evolution and evolution trend prediction were analyzed. The water surface profile of a river was calculated using the energy equation of a constant non-uniform gradient flow equation. The second-order RungeKutta method of a two-dimensional model was used to simulate the impact of construction projects on the flow field during river flood discharge. The results indicate that the Guotan Hub of the Tanghe Navigation Project has a relatively small impact on river flooding so that solved the technical problem of flood impact assessment for Guotan Shipping Hub, which provided reference for relevant departments.

To determine the appropriate ecological flow for small watershed assessment, taking Qingshui River in Pearl River Basin as an example, the monthly ecological flow was calculated by using six hydrological methods such as Tennant method, Texas method and frequency curve method. Furthermore, combined with hydrological rhythm and ecological flow guarantee rate, the advantages and disadvantages of the six methods were compared and analyzed in the dry and flood periods, and the most suitable ecological flow calculation method for small watershed was obtained to determine the ecological flow. The results show that the ecological flow of Hongqi Reservoir section is between 0.12-1.52 m³/s, and the ecological flow of Puzhehei section is between 0.30-4.95 m³/s. The fitting coefficients of hydrological rhythm are 0.993 and 0.998, respectively. The ecological flow guarantee rate is at a good level, and the ecological flow selection result is reasonable. The results can provide a reference for similar national small watersheds to select appropriate calculation method to determine the ecological flow.

Affected by global climate change and human activities, extreme weather events such as drought and rainfall have led to landslides and other disasters in the reservoir dam. At the same time, the impact of evaporation and rainfall on the seepage and stability of the dam slope has not been considered in the design specifications for roller compacted earth-rockfill dams. Taking an earth-rockfill dam as the research object, a finite element calculation model was established to study the impacts of evaporation, rainfall and reservoir water level changes on the pore water pressure, saturation line and safety factor of anti-sliding stability of the earth-rockfill dam. And the cause mechanism of landslide was analyzed in combination with geological exploration and monitoring data. The results show that evaporation and rainfall have a great impact on the pore water pressure of the soil within 0~2 m of the downstream dam slope. The difference between the measured water level of the piezometer and the saturation line calculated by the finite element method is less than 0.4 m. At the initial stage, evaporation improves the stability of the dam slope, but evaporation causes the surface fill of the dam body to shrink and crack continuously, providing a channel for rainfall infiltration. With the increase of the number of rainfall and evaporation cycles and the rising of the reservoir water level, the cracks continue to increase, and the local stagnant areas near the bottom of the cracks are interconnected to form a saturation zone, which leads to the instability and damage of the shallow dam slope downstream of the dam. The research results provide a scientific basis for improving the design specifications of earth-rockfill dams and the early warning of reservoir dam landslides.

In the construction of hydropower projects under the same dam site, the height of water-retaining structures is a key index affected by many factors. In this paper, the responses of three design dam height schemes of a water-retaining structure under static and dynamic superposition conditions were analyzed. The results show that although the dead weight and reservoir water pressure of the arch dam increase with the increase of the dam height, the ground motion response of the arch dam does not all increase with the increase of the dam height. In this calculation case, the beam compressive stress and arch compressive stress on the upstream surface of the arch dam increase with the increase of the dam height, the beam compressive stress on the downstream surface decreases with the increase of the dam height, and the arch compressive stress first decreases and then increases with the increase of the dam height, the forward displacement and openness first increase and then decrease with the increase of the dam height. The research results can provide a reference for the design of water-retaining structures.

Support Vector Machine (SVM) has advantages in small sample simulation prediction, but there is subjectivity in the selection of penalty factor C and kernel function parameter γ in SVM. Therefore, the Harris Hawks Optimization (HHO) algorithm was used to optimize C and γ in the SVM. And then the HHO-SVM mode was established to predict water quality in the Xiyuan tunnel section of Lake Dianchi Caohai. The results show that the prediction accuracy of the water quality prediction model based on HHO-SVM is higher than that of the SVM based on genetic algorithm (GASVM) and the SVM based on whale optimization algorithm (WOA-SVM). It is proved that the HHO is feasible to optimize the parameters in SVM, and HHO-SVM can be used in water quality prediction.

In order to understand the impact of vegetation groups on the hydraulic characteristics of natural meandering river, this paper designed two kinds of layout forms in which the cluster layout simulated the natural cluster vegetation group, and the uniform layout simulated the artificial regular planting vegetation group. Through the flume experiment, the impact of different layout forms of vegetation on the distribution of flow field, hydrodynamic axis, turbulent kinetic energy and transverse circulation structure in continuous bends was explored. The results show that the flow field in the bend appears obvious velocity partition in which the high velocity area is close to the convex bank, and the low velocity area is attached to the concave bank. Vegetation layout makes the distribution range of high and low flow rate areas have the opposite change rule: the high flow rate area increases, while the low flow rate area decreases. Under the action of the bend, the hydrodynamic axis 'gradually swings to the convex bank on the upstream side of the bend top, close to the convex bank near the bend top and swings to the concave bank on the downstream side of the bend top, completing a cycle of motion changes. Vegetation layout makes it swing to the middle of the flume and the deviation degree of cluster layout is greater than that of uniform layout. Under the action of the bend, the distribution of turbulent kinetic energy of water flow is small on both sides and large in the middle. On the whole, the layout of vegetation increases the turbulent kinetic energy on the downstream side of the bent top, and the impact of cluster layout is greater than the uniform. The vegetation layout changes the circulation structure of the section, which is manifested by the swing of the vortex core position and the change of the distribution range.

The drought in the Fenhe River Basin has the characteristics of serious disaster and wide range of influence, which has a great blocking effect on the development of local society economy, so it is of great significance to accurately evaluate the drought situation in the basin. Based on Standardized Precipitation Index (SPI), Standardized Precipitation Evapotranspiration Index (SPEI) and Standardized Runoff Index (SRI), the data were grouped with Tyson polygons. A new comprehensive drought index Comprehensive Index (CI) was constructed by Caussian Copula function. Mann-Kendall trend test was used to test the trend of CI annual series, and the run-length theory was used to extract drought features. It is found that CI can respond well to both hydrological and meteorological droughts. The number of M-K trend series of CI sequence is -3.78. There is a trend towards drought in the upper reaches of the Fenhe River. The intensity of drought in 1970 is the smallest, which is 0.06. Four years from 2013 to 2016 appeared continuous drought, and the maximum intensity is 2.66.

Based on the monthly data of precipitation and average temperature from 36 reference meteorological stations in the six basins in Yunnan Province, 11 climate models published by CMIP6 were used to simulate and predict the change process of meteorological elements in six basins. Three model integration technologies were used to improve the simulation ability and evaluate the suitability of each model and integration method. The optimal method was chosen to predict the possible trend of climate change in six basins under the scenario of SSP2-4.5 in the future. The results showed that precipitation and temperature would increase gradually in the near, middle and long term in the six basins of Yunnan Province. Precipitation might increase significantly after 2060, and temperature might increase significantly after 2050.

In view of the case that the high wave wall changes from inclined to vertical near the dam slope and crest, numerical simulation and physical model tests were used to mutually verify the calculation of wave run-up. The working conditions of different wave elements and water depth in front of the dam were analyzed to study the impact of high wave walls on wave run-up in the reservoir. The results show that when the same wave factor is applied to different reservoir depths, the wave run-up shows an inverted parabolic trend with the reservoir depth; The maximum wave run-up occurs at the intersection of the vertical section of the wave wall and the slope section; When the water depth in front of the wall is higher than the intersection of the vertical section and the slope section, the wave run-up calculated by the standard method is larger compared with the results of mathematical and physical models. The finding can provide a reference for the refined design of the top elevation of high wave walls considering the impact of wave run-up in the design of earth-rockfill dam.