Aiming at the problem of campus waterlogging in Hefei City, this paper constructs a 1D-2D coupled model based on Mike Flood. Firstly, GIS was used to obtain the distribution information of the five types of land surface in the study area by analyzing satellite images, and to automatically divide the catchment areas. Secondly, combined with the measured elevations by using the RTK device, the 3 m×3 m high-precision DEM data of the study area were obtained. A 1D pipeline model and a 2D surface model in the study area were constructed using Mike Urban and Mike 21 respectively, and the two models were coupled on Mike Flood. At the same time, a flowmeter, a liquid level meter and a rain gauge were installed in the campus. The model was calibrated by two measured rainfall events according to R² and E_NSE. Finally, the Chicago rainfall pattern was used to design rainfall scenarios for the 5, 10, 30 and 50 years of recurrence periods, and the severity, inundation range and inundation depth of nodes and pipelines in different rainfall processes were simulated. The simulation results show that main causes for the waterlogging in the campus are due to unreasonable sub-catchment designs, which resulted in limited absorption of rain peak runoff and serious overload of the pipe network.

To improve the state detection effect of large supercritical generator rotor set, the RSO (Cyclic Periodic Electrical Impulse) detection method of large supercritical generator rotor set based on time difference calculation is designed. According to the principle of RSO detection method, the characteristics of time difference pulse signal are extracted. The detection model of coil short circuit is constructed according to the time difference characteristics of signal. Wavelet transform is used to denoise the time difference signal. Fix the rotor position and winding parameters are fixed to determine the fault position of the rotor set. Thus, the RSO detection of the rotor set of large supercritical generator unit is realized. The experimental results show that the difference between the detection results of the power spectrum amplitude of the initial rotor position, the rotor position at startup, the normal broken bar and the broken bar fault is small, which verifies the effectiveness of the proposed method.

As an essential influencing factor of permissible pollution bearing capacity, hydrological condition was changed significantly due to hydropower cascade development. According to the water functional division and water quality control target in Panzhihua reach of Jinsha River, the variation of dynamic permissible pollution bearing capacity (COD, NH3-N) with hydropower cascade development were calculated depending on 2-D hydrodynamic and water quality models. Furthermore, response relationship between permissible pollution bearing capacity and hydrological factors was established to evaluate the effect of hydropower cascade. The results show that the dynamic permissible pollution bearing capacity of COD and NH3-N decreased by 5.2%-27.8% and 3.8%-41.2%, respectively, and its decreased range was found to be more evident in dry season. The variation trend of permissible pollution bearing capacity and near-bank discharge was similar for different conditions. Simple coefficients between the variation of near-bank discharge and permissible pollution bearing capacity were calculated. However, a significant correlation was found with permissible pollution bearing capacity of NH3-N only.

At present, most leakage location methods rely on the hydraulic model of the pipe network. However, when the accuracy of the hydraulic model cannot meet certain requirements, or when the model cannot be established due to the lack of basic data, the model-based method will fail. For this purpose, the study of the leakage location of the water supply network was carried out on the basis of pressure monitoring data. Based on the modeling theory of graph theory, the interpolation method for estimating the node pressure of pipe network was obtained. The leakage position was estimated by analyzing the residual of the pressure monitoring value of each node and the measured value after the leakage occurs. With the help of Bayesian theorem, time-series reasoning was carried out on the positioning results, and the node with the largest probability in a certain period of time was regarded as the location where leakage occurs. By using the case of L Town, the state of pipe network was simulated when single point leakage occured, and the feasibility and location performance of the leakage location method were verified.

Based on the understanding of the concept of happiness river, taking 6 rivers with moderate river scale in Qiqihar Areas as an example, the TOPSIS evaluation model is constructed by combining the proportional scale AHP and AEM to carry out the evaluation on the happiness river, whose goal was divided into 5 guiding layers and 20 indicator layers. The evaluation results show that the overall happiness of the rivers in Qiqihar Areas is grade Ⅲ, which the happiness of Yalu River is grade Ⅳ, the happiness of Alen River, Nuomin River, Yin River and Runjin River is grade Ⅲ, and the happiness of Chaoer River is grade Ⅱ. In addition, the common problems of poor evaluation grade in the results were analyzed and the treatment suggestions were put forward.

Aiming at the problem that the traditional water quality prediction method is difficult to capture the spatial and temporal characteristics of the sample, this paper proposes to establish a CNN-EA-ConvLSTM based water quality prediction model. The convolutional neural network (CNN) was used to reduce the dimensionality of the data and extract the sample features. Then the hidden information among samples was explored by the external attention mechanism. The convolutional long and short-term memory network (ConvLSTM) was further used to capture the spatial characteristics of the data. To achieve optimal results of the model, a genetic algorithm was used to optimize the parameters of the model. The water quality test data of Qinghai Province was used as a sample to simulate and validate the model. The results show that the mean absolute error (M_MAE) of the model is 0.063, the root mean square error (R_RMSE) is 0.012, and the mean absolute percentage error is 2.6%, which are respectively reduced by 18% and 24%, 14% and 25%, 16% and 21% compared with the CNN-EA model and CNN-LSTM model. Therefore, the model can effectively obtain the spatial and temporal characteristics of water quality, attenuate the influence of different samples, and achieve the ideal prediction effect.

In view of the serious problem of scouring under the sluice gate of the Tidal River, taking the Shanghai area as an example, from the perspective of operation and maintenance management, the scouring mechanism under the sluice and the optimization method of operation and maintenance management were explored. Through the flow state analysis and hydraulic jump formula derivation, the variation law among the water level, the critical water depth of the outer river and Froude number in front of hydraulic jump was revealed. The water level solution formula suitable for the operation and maintenance management of the tidal estuary gate was proposed. The results show that the water level of the outer river fluctuates greatly because the tidal estuary sluice is affected by tidal action; When the water level of the outer river is lower than the critical water depth, it should be prohibited to open the gate; When the Froude number in front of the hydraulic jump section of the outer river level is between 4.5 and 9.0, the drainage of the gate can be effectively exerted, and the erosion under the gate can be effectively reduced. The results can provide a reference for the operation and maintenance management of similar tidal estuarine gates.

In the case that hydrological data only include rainfall station data and measured water level data of river monitoring section, it is always a major challenge to improve the accuracy of flood forecasting in small and medium-sized basins. In this paper, flood automatic coding was realized according to data characteristics, and historical floods were classified by using decision tree model, thus the candidate flood groups and eliminated flood groups were obtained and the calculation efficiency was improved. On this basis, a flood forecasting model was established based on the similarity measurement of spatio-temporal hydrological data. Taking the hydrological data of Dafenshui Basin in Jiangxi Province as an example, a case analysis of hydrological data was carried out, and four floods were randomly selected for verification. The results show that the qualification rate of flood peak water level is 100%, and the qualification rate of flood peak time is 75%, the accuracy is high, which has important theoretical value and practical significance in hydrological data research.

In order to explore the microcrack initiation, propagation and damage evolution of rock samples with different crack types (opening and closed), inclination and crack length, based on PFC2D, the fractured rock sample model was constructed and the uniaxial compression numerical tests were carried out. The initial crack and failure characteristics of rock sample under different crack obliquity and type were explained from the perspective of mesoscopics. The damage analysis of rock samples was carried out from the perspective of energy. The results are as follows. With the increase of crack inclination angle, the peak strength of the open type is positively correlated, while that of the closed type decreases first and then increases. The elastic modulus is consistent with the change of peak strength. When 30°≤α≤60°, the microcrack initiation characteristics of the two kinds of fractured rock samples are consistent. When 60°≤α≤90°, the failure paths of the two rock samples are similar. The damage variable and correction coefficient were defined based on the energy of cementation failure, and the evolution model was established. The rationality of the model was verified from the perspectives of fracture type, inclination angle and length. The research results can provide a theoretical basis for further understanding the damage evolution process of fractured rock mass after loading.

Scientific evaluation of flood resource utilization potential is an important prerequisite for carrying out flood resource utilization in the basin and alleviating the contradiction between supply and demand of water resources. To evaluate the potential of flood resource utilization in different sections of the Xizhijiang River Basin, the SWAT model was constructed to simulate the natural runoff process of Xizhijiang River Basin, and the flood resources of the basin were quantified by combining the limit analysis theory. The potential of flood resources utilization in the upper, middle and lower reaches of the basin was evaluated, and the feasibility of flood resources utilization in Xizhijiang River Basin was discussed based on the water quality conditions. The results show that the SWAT model has good applicability in runoff simulation of Xizhijiang River Basin; The inter-annual distribution of flood resources is uneven, and there are great differences in inter-annual flood resources utilization potentiality. The utilization condition of flood resources in the lower reaches is better than that in the upper reaches, while the utilization rate of flood resources in the lower reaches is lower. The average annual current and theoretical utilization potential of flood resources in the lower reaches are 3.48×10⁸m³ and 7.97×10⁸m³, respectively. The water quality indexes of the main stream of Xizhijiang River Basin are not lower than the class Ⅲ standard, and the utilization of flood resources in Xizhijiang River is feasible. This study can provide support for the optimal allocation of flood resources in Xizhijiang River Basin, and it has an important reference value for achievement of high quality development of the basin.