The ultra-short-term reservoir water balance simulation of channel-type reservoirs is the key to the operation of the reservoir, but the results easily appear sawtooth oscillation. The calculation error is due to the fluctuation of the water level in front of the dam by extending the simulation time scale. Among the factors that cause the fluctuation of the water level in front of the dam, taking the load change as an example, the calculation error of water balance caused by the load change was corrected. An ultra-short-term water balance method for channel-type reservoirs was put forward based on one-dimensional hydrodynamics. The method established a one-dimensional hydrodynamic model of the river-reservoir area, and procured the accurate storage capacity by calculating the average water level instead of the water level in front of the dam, thereby eliminating the jagged error. The calculation results show that the simulation accuracy can be effectively improved after eliminating the influence of load changes. The ultra-short-term water balance method for channel-type reservoirs based on one-dimensional hydrodynamics can eliminate the sawtooth error and achieve high simulation accuracy.

In order to realize the prediction of industrial water consumption and improve the prediction accuracy, a mixed strategy was introduced to optimize the sparrow search algorithm (SSA) for improving the global search ability. The improved sparrow search algorithm (ISSA) was used to optimize the parameters of support vector machine (SVM). A support vector machine model (ISSA-SVM) based on hybrid strategy and ISSA was established, and the prediction of industrial water consumption in Ningxia was taken as an example. The results show that the ISSA-SVM model has the characteristics of fast optimization speed and high precision, and it has good applicability and feasibility in industrial water consumption prediction.

The selection of a suitable, efficient and accurate numerical model for groundwater flow is an important basis for numerical simulation work. The analytical solution of water level of seepage calculations of fully penetrating well in a phreatic aquifer was compared with the numerical solution of water level of structured conventional model, structured encrypted model and two unstructured encrypted models to analyze the degree of fit and error. The results show that the accuracy of the numerical simulation of all four models is quite good, and the average absolute error is less than 0.1, R² is greater than 0.98. Compared with the structured traditional model, the accuracy of the structured encryption model is the best, the unstructured 4-level encryption model is the second and the 3-level encryption model is the worst; The accuracy of the unstructured four-level encryption model is slightly worse than that of the structured encryption model, but the number of mesh is reduced by 90%. It is a highly applicable, efficient and accurate numerical model for groundwater flow, which improves the computational efficiency and makes up for the poor applicability of multi-point encryption of the structured encryption model.

To scientifically evaluate the eutrophication degree of water bodies, based on the comparison and analysis of various eutrophication evaluation methods, convolutional neural network (CNN) was introduced to establish the convolutional-eutrophication (CNN-E) model. Based on the monthly-scale water quality and algae monitoring data of Honghu Lake from 2014 to 2019, the comprehensive nutrient index method, BP neural network method and CNN-E model were used to evaluate its eutrophication degree. The mean absolute error, root mean square error, coefficient of determination and Nash-Sutcliffe efficiency coefficient were used to evaluate the performance of the neural network model. The results show that Hong Lake was in a mild eutrophic state for a long time, and the eutrophication level was increasing. In terms of model performance, the four evaluation indexes of CNN-E model are better than BP neural network 0.166, 0.098, 0.078 and 0.087, respectively. The CNN-E model can provide technical support for the prevention and comprehensive management of eutrophication in lake water bodies.

A typical urban channelized (Urban) channel in Liangtan River, Chongqing, was reshaped into six types of channel by a numerical simulation tool named RiverBuilder. And then a two-dimensional hydrodynamic convection-diffusion model was constructed to study the effect of channel morphology reconfiguration on migration and diffusion index, such as water turn-over time (T_TOT), pollutant concentration curve (C_CC), pollutant reaching maximum time (M_MT) and pollutant arriving time (A_AT). The results show that the channel morphology reconfigured by changing the width (W_bf), depth (D_bf), and meandering (M_d) of the Urban channel can inhibit the migration and diffusion of pollutants to a certain extent, but the influencing effect is not as good as that of the composite channel based on the variable D_bf. Meanwhile, the near-natural (Natural) channel has the strongest anti-pollute capacity and inhibiting ability of pollutants diffusion, indicating it is more suitable for the self-purification process of pollutants. It is confirmed that the channel morphology based on disordered and complex changes in W_bf, D_bf, and M_d is closer to the Natural channel, which can provide good eco-hydraulic conditions for the improvement of river water quality.

In order to fully reflect the natural environmental conditions and effects formed and maintained by the river ecosystem, which is the basis of human life, after in-depth analysis of the river ecosystem service function, the calculation method of carbon fixation and other functions was optimized. Based on the functional value method, a scientific and reasonable river ecosystem service value (ESV) evaluation methods covering more comprehensive functions was proposed, and Nanhe River ESV in Liyang City was studied with an example. The total ESV of Nanhe River in 2019 is 1.242 billion yuan, with the value of regulation, support, supply and cultural functions accounting for 76.5%, 12.2%, 11.1% and 0.2% respectively, which can provide reference for natural resource asset accounting and ecological compensation standards.

In order to maximize the comprehensive benefits of cascade reservoirs in the lower reaches of the Jinsha River, it is necessary to study the application mode of joint flood control scheduling. Adopting the flood process in different areas with the incoming water from the Jinsha River and the Minjiang River as the mainstay, the flood regulation calculation, flood control analysis, flood control storage capacity utilization, and cascade power generation benefit calculation of the cascade reservoirs in the downstream of the Jinsha River were studied by using relevant mathematical analysis methods. The flood control distribution method of the cascade reservoirs in the lower reaches of the Jinsha River to meet the goal of flood prevention in the Chuanjiang River was studied. The research results show that on the premise of meeting the flood control requirements of the downstream Chuanjiang River, from the perspective of making full use of flood resources and increasing the overall power generation benefits of cascade power stations, in the distribution and dispatching of flood control, it is recommended that the order of flood control storage capacity utilization of cascade reservoirs downstream of the Jinsha River is as follows: Wudongde Reservoir, Xiluodu Reservoir, Baihetan Reservoir. The research results can provide references for rational and scientific flood control operation of cascade reservoirs in the lower reaches of the Jinsha River and similar basins.

As an important part of rural energy, small hydropower plays an important role in driving rural economic development and promoting energy saving and emissions reduction. However, there is a lack of scientific guidance in the process of optimal operation of small hydropower stations in most areas. This paper takes the Mengdong River basin as an example to establish a cascade optimal operation model of small hydropower stations based on runoff simulation. The Xinanjiang (XAJ) model is combined with Manning's formula to establish a semi-distributed hydrological model for simulating the influence of reservoir operation on runoff. The successive approximation dynamic programming method is used to achieve optimal operation by considering the actual operating characteristics of small hydropower stations. The results demonstrate that the semi-distributed hydrological model has a good performance in the river basin. The changing pattern of simulated and measured runoff processes is basically consistent, and the Nash-Sutcliffe efficiency coefficient and coefficient of determination between the measured and simulated series are 0.86. The optimal operation model can reduce the spillage water effectively, which improves the utilization efficiency of water energy. Meanwhile, the optimal operation can match the actual intra-day features of small hydropower, indicating its feasibility in applications.

Under the situation of "abundant and dry electricity supply" in Sichuan Province, affected by complex market environment factors and intensified power supply competition, accurate prediction of power generation plan is the key to optimizing the power generation mode of hydropower in dry periods in the basin. This paper analyzed the power supply capacity of the main grid, new energy power, power regulation capacity of the national power grid, outsourcing power transmission capacity and power generation capacity of the provincial hydropower and thermal power, respectively, and established a forecasting method for the generation plan of the cascade hydropower stations in the Dadu River basin. The results show that the prediction accuracy of power supply in main grid is 91.8%, and the prediction accuracy of wind and solar power generation is 87.4%. The prediction accuracy of the medium and long-term overall power generation plan of the hydropower station belonging to Guoneng Dadu River Basin Hydropower Development Co., Ltd. is 90.5%. The model can effectively predict the electricity quantity of each power source and calculate the medium and long-term power generation plan of the cascade hydropower stations in the Dadu River basin, and intelligently generate the fluctuation process under different working conditions based on the prediction results, which provides technical support for the scientific dispatching and operation of the cascade hydropower stations.

The multi-energy complementary balance model used in the demonstration of the scale of pumped storage in the new power system only considers the high cost of deep peak shaving brought about by the minimum technical output of thermal power for coal-fired unit peak shaving, ignoring the economic issue of coordinated operation between coal-fired and pumped storage units. This paper proposed to use the economic output coefficient of thermal power and the minimum output coefficient of thermal power as discriminant indicators for coordinated operation of thermal power and pumped storage, and incorporate it into the original multi-energy complementary balance analysis model. From an economic perspective, the scale of pumped storage was predicted with the minimum total annual cost as goal. Taking a certain power grid as an example, the process of demonstrating the scale of pumped storage energy was illustrated, which provides reference for similar power grid development planning and early engineering design work.