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.

Seepage analysis has always been the focus of channel safety and stability research. Establishing a diagnosis method based on osmotic pressure monitoring data and combining the detection information to achieve osmotic safety is also an effective way to ensure the safety and stability of osmotic engineering. In this paper, a high-fill channel section of a water diversion project is taken as an example, and the study of osmotic safety is carried out by analyzing the engineering characteristics of linear engineering. High-fill channel sections were selected to study the temporal and spatial characteristics of various monitoring data reflecting the osmotic behavior of linear engineering; that is, these data reflect the time-varying regularity characteristics of the osmotic pressure value and the changing regularity of environmental variables. A single-point multifactor model of the monitoring data was established by establishing an evaluation index system, combining the monitoring index value method and the cloud model theory method according to the distribution law of the measured data and considering the uncertainty of the osmotic pressure data. Additionally, this model was integrated with the set pair analysis method to determine the monitoring data evaluation level; channel detection data information was collected, the abnormal detection of detection information was realized, and the monitoring data results were used to verify the detection results. In this way, an adaptive evaluation method reflecting the working behavior of high-filled channel sections is established, and a diagnostic technology for the safe operation of high-filled channel sections of linear engineering is proposed. The application results show that this method is suitable for engineering an osmotic safety assessment.