Most existing parameter inversion methods of the dynamic constitutive model parameters do not take the concept of uncertainty into account. Therefore, the adaptive cloud transformation algorithm (AGCT) was combined with RBF neural network (RBFNN) to construct an adaptive cloud neural network parameter inversion model (AGCTNN), which converts the uncertainty concept into quantitative values and better takes into account the influence of the randomness and ambiguity between dam systems on the inversion of dynamic parameters. AGCT was compared and analyzed with three traditional clustering algorithms, K-Means, SOM and DBSCAN, to verify the superiority and feasibility of the algorithms. The inversion analysis was then carried out on engineering examples using two inversion models, AGCTNN and RBFNN. The results show that the positive coupling results of the proposed inversion model are in better agreement with the measured values, and the error range between the measured and inverse values of peak acceleration at measurement points is reduced from 8.73%-25.17% to 2.31%-8.16%, which confirms the reasonableness of the inversion model and the possibility of its application in practical engineering.

The structural joints of face rockfill dam are the most important part and the weakest link of the seepage control system of dams. Taking a face rockfill dam as an example, the position, width, length of vertical failure joints and permeability coefficient of cushion area were simulated by finite element method. And then the seepage field law and permeability stability after water stop failure were studied. The results show that in the case of single fracture failure, the position, width, length of fracture and permeability coefficient of cushion are positively correlated with control factors of seepage flow, and the length of vertical fracture and permeability coefficient of cushion are the most significant. When multiple joints fail simultaneously, the seepage flow at cracked panel can be calculated by summing the failure results of single joints, and the results are safety. At the same time, it is suggested that the permeability coefficient of cushion should be 1×10^-5m/s or slightly smaller. The results can provide reference for the design and construction of similar projects.

Aiming at the large-scale backflow problem in the forebay with large forward diffusion angle of the pumping station, based on the CFD software, the RNG κ-ε model was used to simulate the flow state of the forebay of a pumping station. The hydraulic characteristics of the forebay under the original scheme were analyzed. The rectification effect of the figure-eight-shaped diversion piers was explored. The results show that adding figure-eight diversion piers in the forebay could effectively eliminate the large-scale backflow on both sides. The backflow area of the plane flow field disappeared, and the uniformity of the axial flow velocity distribution of the side unit was improved by 11.03% and 11.32% compared with the original scheme. The research results can provide a reference for the rectification of the forward forebay pumping station.

When the water conveyance tunnel in long-distance water diversion project passes through the potential earthquake area, there is a major engineering safety hazard caused by active fault dislocation. It is of great engineering significance to reveal the influence of fault dislocation on the stability of water conveyance tunnel. Based on ABAQUS finite element analysis software, considering the complex dynamic interaction characteristics of surrounding rock, fault, lining and internal water, the three-dimensional numerical simulation of surrounding rock-fault fracture zone-lining-internal water system was carried out, and the influence of different fault dislocation momentum, width and cohesion on the stability of water conveyance tunnel was analyzed in depth. The results show that the vertical displacement and damage degree of the lining structure of the water conveyance tunnel decrease under the conditions of small fault dislocation, fault width and large fault cohesion, which has a certain reference value for the theoretical study of the dynamic response of the water conveyance tunnel across faults.

Roughness is an important parameter which reflects water transport capacity of water diversion project. The roughness value can be calculated by observing the frictional head loss. However, how to estimate the roughness is always a difficult problem in hydraulics for the new unworked water diversion project. The surface roughness of water conveyance buildings is a vital factor affecting the roughness value. So, this study proposed a surface roughness measurement method relies on hierarchical sampling. Taking a new in-situ reinforced concrete lining tunnel as an example, the statistical distribution of surface roughness and the minimum sample size of tunnel roughness estimation were analyzed. On this basis, the roughness value range of the tunnel project was estimated, which has a certain reference significance for predicting the water transport capacity of the new project and inspecting the construction quality of the lining.

As a diversion tunnel, the pressure distribution of the tunnel body, the top and the bottom of the tunnel is very important to avoid cavitation, vibration and other undesirable phenomena. Taking the diversion tunnel of Yebatan Hydropower Station as the research object, this paper simulated the pressure distribution of the diversion tunnel based on the physical model test, as well as the three-dimensional numerical model established by the RNG k-ε turbulence model and the VOF method. The reliability of the three-dimensional numerical simulation was verified by the experimental data. The results show that the main flow of the diversion tunnel, the pressure distribution of the tunnel body at the sudden change of hydraulic boundary such as the inlet chamber section, the turning section and the plug section, as well as the variation trend of the bottom and top pressure along the way were analyzed. The research results show that the pressure distribution along the bottom and top of the diversion tunnel shows a downward trend, but at the sudden change of hydraulic boundary such as the sluice chamber section, the plug section and the turning section, the pressure value will show a large fluctuation of rapid increase or decrease.

sedimentation basin is one of the important water conservancy facilities to reduce the sand content of water flow in agricultural irrigation projects, which plays an important role in reducing the sand content of water flow and improving the utilization rate of water flow. In order to improve the efficiency of sand sedimentation basin discharge, a linear sand collection culvert structure was designed based on the S-type structure of the sedimentation basin in the Weigan River dry canal. The hydraulic characteristics of the S-type and linear sand collection culvert water flow were clarified by combining the experimental results of prototype observation with physical model tests and numerical simulations. The experimental results show that Fluent software can effectively simulate the operation process of local sand traps; Due to the low flow velocity at the end of the S-type sand trap, the sediment accumulation in the sand trap is caused, while the linear structure effectively increases the overall flow velocity in the sand trap, and the flow velocity in the corridor at the end of the sand trap is increased from 0 m/s to 0.4 m/s-0.7 m/s, thus improving the sand discharge efficiency of the sand trap. This study can provide scientific basis and technical support for optimizing the structure of the sand sink and improving the sand discharge efficiency of the sedimentation basin.

Flexible mattress is a beach protection measure which is widely used in the waterway regulation projects in the middle and lower reaches of the Yangtze River. The in-situ service condition of the flexible mattress has a great impact on the protection efficiency for the beach and shoal. Aiming at the difficulty of identifying and quantifying the deformation of the bulge of the flexible mattress, a model test was carried out to study the strain response characteristics of the bulge deformation. The result shows that the strain under bulging deformation presenting a parabolic strain distribution characteristic on the cross-section, and the area with the largest strain is located on the top of the bulging body. Based on the strain distribution characteristics of bulge deformation, the shape prediction formula was proposed. Furthermore, the bulge deformation identification method and shape inversion method were put forward. The results can provide the reference for the deformation monitoring and assessment of the flexible mattress.

A static dynamic triaxial instrument was used to carry out concrete triaxial dynamic mechanical tests by setting different strain rates (10^-5/s, 10^-4/s, 10^-3/s, 10^-2/s) and different lateral confining pressures (4.5 MPa, 6.0 MPa, 9.0 MPa and 12 MPa). The evolution law of the total energy density, elastic strain energy density and dissipated energy density of concrete with axial strain under dynamic triaxial compression was studied. The influence of lateral confining pressure and strain rate on the energy evolution law of concrete was analyzed. The results show that with the increase of axial deformation, the total energy density and dissipated energy density of concrete increase nonlinearly, and the elastic strain energy density first increases and then decreases. The increase of strain rate and lateral confining pressure improves the specific energy storage limit of concrete, and the energy storage limit of concrete shows a strong rate sensitivity; The relationship of energy storage limit and the lateral confining pressure is approximately linear; With increase of the lateral confining pressure, the rate sensitivity of the concrete energy storage limit shows a decreasing trend.

The deep learning technology was used to study a method of predicting the performance of SCC based on the mixture image information during the mixing process. Twenty-five sets of videos of the SCC mixing process with different performances were recorded. According to the slump flow and T500 measured values and combined with the visual inspection, the SCC mixes were classified into three performances: qualified, insufficient fluidity and segregation. By processing the videos into image sets, the deep learning models were built using image classification and target detection respectively. The models learn and train the image features of the mixes to realize the prediction of SCC performance. The results show that both image classification and target detection methods can achieve more than 98% accuracy on the validation set, which provides a reference for adjusting the mix proportion in real-time and realizing the smart production of SCC.