Numerical simulation and process parameter simulation of dam vibration compaction construction

Introduction: vibration compaction is also essential in building earth-rock dams because it ensures the density of soil, mechanical stability, and avoids deformation with time. Nevertheless, the definition of internal soil behaviour under dynamic stress cannot be done properly by field data alone, and it is impossible to optimise the vibration frequency, lift thickness, and roller speed. Methods: the rationale behind this research is to unite the numerical simulation with the machine-learning-based prediction to evaluate the performance of vibration compaction and identify the most appropriate construction parameters to use in dam embankment projects. The Dam Vibration Compaction Dataset captures realistic field, lab, machine, and simulation-based parameters for earth-rock dam compaction, comprising 25 features. Data Preprocessing entailed normalization of data and filtering of noisy sensor readings. Results: internal stress distribution and densification behavior in soil were simulated using a coupled soil-vibration dynamic model based on the Finite Element Method. Simultaneously, a Scalable Random Vector Machine algorithm, combining the Scalable Support Vector Machine and the Random Forest algorithms, was created to forecast compaction quality under varying parameters. Conclusions: the FEM+SRVM model, implemented using Python, demonstrated a high level of prediction, with the most effective parameters being the vibration frequency of 30 Hz and Root Mean Square Error of 0,0083. The combined numerical and ML method enables a potent means of dam vibration compaction analysis and optimisation to remove the trial-and-error method and increase the reliability and quality of construction.