During the operation period, the deformation of an ultra-high arch dam is affected by the large fluctuation of the reservoir water level. Under the dual coupling of the ultra-high dam and the complex water level conditions, the traditional variational analysis method cannot be sufficiently applied to its deformation analysis. The deformation analysis of the ultra-high arch dam, however, is very important in order to judge the dam safety state. To analyze the deformation law of different parts of an ultra-high arch dam, the panel data clustering theory is used to construct a Spatio-temporal characteristic model of dam deformation. In order to solve the difficult problem of the fluctuating displacement of dam deformation with water level effect, three displacement component indexes (absolute quantity, growing, and fluctuation) are proposed to characterize dam deformation. To further optimize the panel clustering deformation model, the objective weight coefficient of clustering comprehensive distance is calculated based on the CRITIC (CRiteria Importance Through Inter-criteria Correlation) method. The zoning rules of the ultra-high arch dam are established by using the idea of the CSP (Constraint Satisfaction Problem) index, and the complex water level of the reservoir is simulated in the whole process. Finally, the dynamic cluster analysis of dam deformation is realized. Through a case study, three typical working conditions including the rapid rise and fall of water level and the normal operation are calculated, and the deformation laws of different deformation zones are analyzed. The results show that the model can reasonably describe the deformation law of an ultra-high arch dam under different water levels, conveniently and intuitively select representative measuring points and key monitoring parts, effectively reducing the analysis workload of lots of measuring points, and improve the reliability of arch dam deformation analysis.

The seepage of the dam is an important representation of the operation characteristics of the dam, and there are many factors affecting the seepage with a certain lag. It is still difficult to predict its change and sensitivity because of complex operating conditions. At present, the lag-sensitivity of influence factors of the dam seepage has not been studied. The time series influence factors of seepage are determined by HTRT (hydrostatic-thermal-rainfall-time) model in this paper. To avoid the pseudo fitting of conventional methods, HTRT model nested random forest algorithm is used to establish the predicting model of the dam seepage. And MIC algorithm is used to achieve the dual purposes of time lag and sensitivity analysis. Firstly, the time lag of relationship between seepage and its influencing factors is characterized, and the most appropriate lag time of the HTRT model factors is determined. Secondly, independent correlation analysis on all influencing factors is carried out and the sensitivity of each factor is analyzed by MIC. Meanwhile, the sensitivity of the factors to seepage is quantitatively analyzed by the two parameters of %IncMSE and IncNodePurity of RF algorithm. The sensitivity of influencing factors is analyzed by comparing MIC results with RF results. Combined with the case, taking the error of fitting prediction as the evaluation index of seepage prediction, the prediction accuracy of MIC-RF model, RF model and MIC-BPNN (Back Propagation neural network) model is calculated and compared. Case study showed that MIC- RF monitoring model has high prediction accuracy, strong adaptability and high robustness in dam seepage, and the sensitivity and time lag of influencing factors can be quantitatively analyzed. The water pressure and rainfall of the lag time are 14 days and 16 days respectively. The sensitivity study of the time series influencing factors of seepage showed that the water pressure component is the main controlling factor of seepage, and rainfall component is more sensitive to later stage. The MIC-RF model can be used as a new dam seepage safety monitoring model.

The stress behavior of key parts of concrete dams is related to the safe operation of the dam. However, the stress sensors in concrete are susceptible to aging and failure with increasing service life. Estimating the structural stress under sensor failure or data loss scenarios for concrete dams in service is essential and complex. This study presents a stress estimation method driven by the observation data. Firstly, a one-to-one correspondence exists between dam deformation reflecting the load effect and structural stress. Estimating the structural stress by simulating load effects with dam deformation is more convenient when it is hard to simulate complex load effects directly. Therefore, based on the observed data before stress sensor failure, the spatial–temporal relationship between structure stress and multi-point deformations of a concrete dam is developed using convolutional neural networks (CNN) and long short-term memory (LSTM). An improved particle swarm optimization algorithm combined with swarm information entropy (SIE–APSO) is proposed simultaneously for tuning the network’s hyperparameter and accelerating the convergence. Finally, the stress estimation of the target part of the concrete dam in service is obtained. The case shows that it is valid and feasible. The RMSE decreased by approximately 21–58%, MAPE decreased by 19–58%, and ARV decreased by 22–94% compared with the load-stress relationship model.

Modal parameter estimation is crucial in vibration-based damage detection and deserves increased attention and investigation. Concrete arch dams are prone to damage during severe seismic events, leading to alterations in their structural dynamic characteristics and modal parameters, which exhibit specific time-varying properties. This highlights the significance of investigating the evolution of their modal parameters and ensuring their accurate identification. To effectively accomplish the recursive estimation of modal parameters for arch dams, an adaptive recursive subspace (ARS) method with variable forgetting factors was proposed in this study. In the ARS method, the variable forgetting factors were adaptively updated by assessing the change rate of the spatial Euclidean distance of adjacent modal frequency identification values. A numerical simulation of a concrete arch dam under seismic loading was conducted by using ABAQUS software, in which a concrete damaged plasticity (CDP) model was used to simulate the dam body’s constitutive relation, allowing for the assessment of damage development under seismic loading. Utilizing the dynamic responses obtained from the numerical simulation, the ARS method was implemented for the modal parameter recursive estimation of the arch dam. The identification results revealed a decreasing trend in the frequencies of the four initial modes of the arch dam: from an undamaged state characterized by frequencies of 0.910, 1.166, 1.871, and 2.161 Hz to values of 0.895, 1.134, 1.842, and 2.134 Hz, respectively. Concurrently, increases in the damping ratios of these modes were observed, transitioning from 4.44%, 4.28%, 5.42%, and 5.56% to 4.98%, 4.91%, 6.61%, and 6.85%%, respectively. The correlation of the identification results with damage progression validated the effectiveness of the ARS method. This study’s outcomes have substantial theoretical and practical importance, facilitating the immediate comprehension of the dynamic characteristics and operational states of concrete arch dam structures.

The concrete strength criterion is the basis of strength analysis and evaluation under a complex stress state. In this paper, a large number of multiaxial strength tests were carried out, and many mathematical expressions of strength criteria were proposed based on the geometric characteristics and the assumption of a convex function. However, the rationality of the assumption of a convex function limits the use of these strength criteria. In particular, misjudgment will occur near the failure curve surface. Therefore, this paper does not assume the shape function of the criterion in advance. By collecting experimental data and using a machine learning method, it proposes a method of hidden function of failure curve surface. Based on 777 groups of experimental data, the random forest (RF), the back propagation neural network (BP) and the radial basis neural network (RBF) models were used to analyze and verify the feasibility and effectiveness of the method. Subsequently, the results were compared with the Ottosen strength criterion, the Guo Wang strength criterion and the Drucker–Prager (DP) strength criterion. The results show that the consistency between the strength criterion model established by the machine learning algorithm (especially random forest) and the experimental data is higher than the convex function multiaxis strength criterion of the preset failure surface shape. Moreover, the physical significance is clearer, the deficiency of the convex function failure surface hypothesis is avoided and the established multiaxial strength criterion of concrete is more universal.