Affected by hydrodynamic excitation and other factors, the opening-closing operation of hydraulic gates exhibits multi-field coupling effects and complex nonlinear dynamic characteristics, leading to difficulties in identifying equipment safety states. Test data of gate operation demonstrate that artificial neural network algorithms can identify hydrodynamic excitation disease features and accurately predict its development trends. To address this, BP and GA-BP neural networks were employed to construct identification and prediction models for hydrodynamic excitation disease. These models were applied to identify and forecast the effective values of reel vibration, with model performance evaluated using metrics including Relative Error (R_RE), Mean Absolute Percentage Error (M_MAPE), and Root Mean Square Error (R_RMSE). Compared to the BP model, the results indicate that the GA-BP model achieves reductions of 20.77% in R_RE, 4.74% in M_MAPE, and 6.27% in R_RMSE, demonstrating superior fitting to measured samples and enhanced stability with extended prediction durations, thus providing critical technical support for engineering risk mitigation and hazard prevention.

To address the issue of reduced detection accuracy under complex working conditions due to the fixed threshold of the isolation forest algorithm, an anomaly detection method for ship lock miter gate monitoring data based on singular spectrum analysis (SSA) and an improved isolation forest (KMIF) is proposed. The SSA is employed to decompose and reconstruct the monitoring data, and separate the trend and noise components. The isolation forest algorithm is improved by incorporating K-Means++ clustering to dynamically set anomaly thresholds for different monitoring datasets. The noise component is then fed into the improved isolation forest algorithm for training and anomaly detection. Taking the stress and vibration data from multiple measuring points of the lower lock miter gate in Jiangsu ship gate project as an example for validation, the results show that the proposed SSA-KMIF method performs excellently in terms of false positive rate, precision, recall ratio, and accuracy. It demonstrates high accuracy and flexibility, which provides a reliable technical support for health monitoring of ship lock miter gates.

To improve the efficiency and accuracy of fault diagnosis for hydroelectric units, combination of multifractal detrended fluctuation analysis algorithm and probabilistic neural network was used to establish a vibration signal feature extraction and recognition model. The binary gravity search algorithm was used to optimize its parameters. The results show that the classification accuracy of the feature extraction and recognition classification model can be improved to 99% and reduce the signal processing time to about 1.3 seconds after optimizing by the binary gravity search algorithm. The proposed vibration signal feature extraction and recognition model for hydroelectric units can significantly distinguish between the normal working state and the fault working state of hydroelectric units, achieving the purpose of using vibration signal features to diagnose faults in hydroelectric units.

The pumped storage power stations are critical infrastructure for achieving carbon neutrality goals. Based on the three-dimensional computational fluid dynamics method, the head loss characteristics in the combined diversion shaft-surge chamber arrangement are investigated. Firstly, the three-dimensional model from the upstream inlet to the inlet of the unit is established. Then, the hydraulic characteristics under different arrangement types are analyzed. Finally, the influence of the diameter of the turning section and the impedance holes on the head loss is explored under the combination arrangement. The results indicate that the pressure distributions are similar in combined and uncombined arrangements. The hydrodynamic characteristics are not deteriorated and reflective water hammer is more effective in combined arrangements. Under the combined arrangement, the larger the diameter of the turning section and the impedance hole, the smaller the head loss coefficient is. This study can provide theoretical references for the design of new structures for pumped storage power stations.

The Longyangxia and Liujiaxia Reservoirs in the upper reaches of the Yellow River have annual regulation capacity and undertake comprehensive utilization tasks such as flood control, water supply and irrigation, and power generation in the Yellow River Basin. The coordination and consistency of multiple objectives need to be achieved by constructing a multi-objective optimized dispatching system for cascade reservoirs. A multi-objective scheduling model has been established for the Longyangxia-Liujiashan cascaded reservoirs, with the goals of maximizing the peak shaving rate, total power generation, and average sediment flushing ratio. The model is solved using the NSGA-Ⅲ algorithm, and an analysis is conducted regarding the competitive relationships among the objectives of flood control, power generation, and sediment flushing. The established multi-objective optimization scheduling scheme is further evaluated through a developed indicator system, and the TOPSIS method is applied to optimize the set of scheduling solutions. The results show that there is a significant competitive relationship between the objectives of power generation and flood control; No significant competition exists between the objectives of sediment discharge and flood control, and there is some competition between the objectives of sediment discharge and power generation. Through a comparison of the optimal scheme and the actual scheduling data, it can be seen that the benefits of flood control, power generation, and sediment discharge in the optimal scheme increased by 20.89%, 16.02%, and 3.61%, respectively, compared to the actual scheduling.

Constructing a high-precision dam settlement prediction model is of great significance for ensuring the safety and risk control of dam during the construction period. Taking dam height, rainfall and aging as the influencing factors of dam settlement deformation during construction period, the long-term and short-term memory neural network LSTM algorithm is introduced, and the attention mechanism is embedded. Thus, a prediction model suitable for dam settlement of concrete face rockfill dam during construction period is proposed. The engineering application shows that the attention-LSTM model makes up for the defect that the LSTM cannot dynamically adjust the weight coefficient at the network layer, improves the computational efficiency and accuracy of the model, and has better nonlinear data processing ability, which can more accurately reflect the change trend of monitoring data in the time dimension during the construction period. The relevant experience can be used as a reference for similar projects.

In order to explore the influence of pressure pulsation of the pumping unit on the powerhouse structure, the powerhouse of Dayuzhang pumping station was taken for an example. Based on the prototype observation data, the vibration source composition and vibration characteristics of powerhouse structure were analyzed using three-dimensional finite element simulation. The safety of the structure was analyzed and evaluated from the perspective of structural resonance check and vibration response. The results show that the hydraulic pulsation caused by RSI and the rotational frequency excitation caused by the operation of the unit have the greatest impact on the vibration of powerhouse under the stable operation condition of the unit, and the natural frequency of the local floor structure has a small degree of coincidence, which is easy to resonate. However, from the perspective of vibration response, the vibration response of each local part is within the allowable range, the outlet elbow and pump seat are the largest, and the pump floor slab is the smallest. This study has important theoretical value and practical significance for realizing the long-term and safe operation of the pumping station powerhouse structure.

There are problems of non-convergence and easy false alarm when formulating dam deformation monitoring indicators (which belong to fixed limits) based on the conventional low-probability method. A calculation method for formulating deformation monitoring indicators based on the low-probability method of separating aging components is proposed. Firstly, the statistical model of dam deformation is established to separate the time-dependent component. Then, aiming at the time series deformation of deducting the aging component, the annual extreme value is selected as the subsample. The corresponding deformation of the annual most unfavorable reservoir water level and temperature is selected as the subsample. The corresponding deformation of the unfavorable water level and temperature based on the combination of orthogonal test method is selected as the subsample. Then the statistical test is carried out, and the small probability method is used to formulate the deformation allowable value of deducting the aging component. Finally, the aging component is superimposed to obtain the non-convergence deformation monitoring index of the dam. Combined with the measured data of a deformation non-convergence gravity dam in southwest China, the analysis shows that compared with the monitoring index proposed by the conventional small probability method, the method based on the separation time component fully considers the time effect and enhances the reliability of the monitoring index.

The pumped storage power station has the characteristics of frequent unit start-up and shutdown and working condition switching, which has great influence on the safe operation of the power station. At present, the on-site monitoring data of the vibration response of the underground powerhouse structure of pumped storage power station under the vibration load of the unit are few, especially the vibration monitoring of the unit under the transient conditions such as unit start-up and shutdown and load rejection. This paper takes the underground powerhouse of a pumped storage power station as the research object, and carries out the dynamic characteristics monitoring analysis under the transient conditions of power generation, pumping switch and different output load rejection. The results show that the peak vibration response of each typical part of the powerhouse structure under transient condition is obviously greater than that under steady condition, and the vibration response under power generation on and off condition is greater than that under pump pumping condition, but the vibration displacement and acceleration can basically meet the recommended limits of the current vibration standard. Under 100% load rejection conditions, the vibration response of the powerhouse structure is the strongest, and the maximum vibration acceleration can reach more than 30 m/s², which is easy to cause impact damage to the powerhouse structure. It is recommended to avoid 100% load rejection conditions during daily operation and maintenance of the power station. In case of occurrence, it is necessary to timely detect the key structural parts of the powerhouse to eliminate safety risks.

It is the key issues of reasonably and accurately predicting the thrust and torque of tunnel boring machines (TBM) to realize the intelligent control of TBMs. This paper proposes a two-stage prediction method of knowledge-data-driven spatio-temporal stacked convolutional network (KD-NTS-GAT). Firstly, based on expert knowledge and the NTS-NOTEARS method, a new information fusion technique is proposed. The discrete expert experience and the continuous NTS-NOTEARS indicators is mapped and smoothly fused through clustering. The causal relationships among the key operating parameters of the TBM is quantitatively extracted to improve the authenticity of the causal relationships significantly. Then, causality is further combined as a prior knowledge with stacked convolutional network deep learning model for predicting thrust and torque of TBM. Taking the bid Ⅳ of Xinjiang Water Conveyance Tunnel Project as an example, a comparative analysis of the KD-NTS-GAT method and the pure data-driven method shows that the KD-NTS-GAT has better prediction capability on thrust and torque. The conclusions can provide a reference for the intelligent control of TBM construction.