The elastic stiffness of spudcan foundations in stiff-over-soft clays exhibits changes similar to “punch-through” failure, creating significant uncertainty for jack-up platform operations. This study conducted a three-dimensional small-strain finite element analysis on this specific topic to discretely simulate the spudcan elastic stiffness profile in stiff-over-soft clay. The influence of the soil surface, layered interface, and their coupling effects were isolated and separately evaluated, and a simple semi-theoretical framework for the influence zone was proposed. The key parameters of layered soil (thickness ratio, shear modulus ratio, soil heterogeneity coefficient, and backflow) affecting the influence mechanism of spudcan elastic stiffness were evaluated and analyzed. It was found that the effects of the soil surface and layered interface competed with each other. The vertical deformation mechanism of the spudcan reduces the “punch-through” failure risk of elastic stiffness by transferring more of the soil deformation to the bottom soft clay layer. Based on the findings from the parameter study, a simplified profile is proposed to predict the variation of the spudcan elastic stiffness. The proposed prediction method provides a comprehensive view of elastic stiffness in stiff-over-soft clay for offshore in-site assessment.

This study introduces the Spatio-Temporal Attention Enhanced Encoder-Decoder Damage Prediction Network (STAE-EDDPNet), an innovative deep learning model designed to enhance the predictive capabilities of coal-rock damage infrared temperature fields, which is crucial for the safe production in rock engineering and mining engineering. STAE-EDDPNet integrates a spatio-temporal attention mechanism, significantly improving the capture of complex nonlinear spatio-temporal information in rock infrared radiation. Compared with baseline models such as 3DCNN, ConvLSTM, and EDDPNet, STAE-EDDPNet demonstrated superior performance in both single-step and multi-step forecasting tasks. Test set results show that its predictive accuracy is 25.56% higher than 3DCNN, 5.69% higher than ConvLSTM, and 0.19% higher than EDDPNet. The study also found that the characteristics of brittle failure rock data significantly affect model training and predictive performance, providing a direction for future data collection and experimental design improvements. The introduction of STAE-EDDPNet not only promotes the application of infrared monitoring technology in the field of safety monitoring but also provides valuable reference for rock damage early warning.

The online identification of rock damage states is crucial for safety monitoring in geotechnical and mining engineering. By analyzing spatiotemporal evolution patterns of infrared radiation in various rock damage states, we established the first infrared temperature field dataset for rock damage state identification. We then constructed a deep convolutional neural network, RESD-CNN, and performed its training and optimization. Results showed that infrared radiation patterns of different rock samples exhibit similarities. RESD-CNN achieved outstanding performance in identifying rock damage states with metrics of ACC 99.04%, Precision 99.39%, Recall 99.52%, and F1-score 99.46% on the validation set. Generalization tests on datasets of different rock types revealed that RESD-CNN significantly outperformed traditional classification methods, demonstrating the feasibility of infrared radiation technology for intelligent coal rock damage identification. This research provides a crucial foundation for developing online identification and early warning systems for rock damage evolution in engineering.