Settlement monitoring and prediction of offshore deepwater breakwater based on PS-InSAR and CNN-LSTM-SE

In the management and maintenance of port area operations, the ability to swiftly and effectively monitor and predict the settlement deformation of breakwaters remains a critical challenge. Addressing the difficulties in monitoring and predicting breakwaters under harsh offshore and deep-water conditions, we propose an innovative method combining PS-InSAR and CNN-LSTM-SE for predictive analysis. This study utilized 88 Sentinel-1A ascending radar satellite images acquired between January 2019 and December 2021, employing PS-InSAR technology to invert and derive deformation values along the radar line of sight in the port area. Focusing on the eastern breakwater, we extracted time-series settlement data from nine monitoring points through data transformation and applied the CNN-LSTM-SE neural network algorithm for predictive analysis, coupled with a risk assessment of breakwater settlement. The results indicate that from 2019 to 2021, the settlement rate of the eastern breakwater ranged from -140 to -20 mm/a, exhibiting a wave-like trend that progressively intensified from the shoreward side to the deep-water side, with a maximum cumulative settlement reaching 356.1 mm. The predictions from the CNN-LSTM-SE model aligned closely with monitoring results, with a correlation coefficient exceeding 0.95. Compared to other methods, CNN-LSTM-SE demonstrated superior predictive accuracy, making it well-suited for settlement forecasting of offshore deep-water breakwaters. High-risk settlement areas in the port are likely to face structural instability due to settlement rates and differential settlement. Specifically, Zone I of the western breakwater reclamation and the eastern breakwater slope are vulnerable to ground settlement and structural damage caused by heavy loads and uneven load distribution, respectively. To mitigate these risks, it is imperative to establish a multi-tiered monitoring and early warning system to capture real-time changes in the foundation. These research findings provide essential technical support and data reference for the safe operation and maintenance of port areas.