Predicting load–displacement of driven PHC pipe piles using stacking ensemble with Pareto optimization

Predicting load–displacement of driven PHC pipe piles using stacking ensemble with Pareto optimization

The utilization of precast pre-stressed centrifugal concrete (PHC) piles is widespread in high-rise construction owing to their robust load-bearing capacity and enduring characteristics. Yet, accurately predicting the load– displacement traits of these piles remains a formidable engineering challenge. This study presents a dependable and user-centric solution for emulating PHC pile load–displacement behavior, focusing on pile diameters ranging from 0.3 to 0.6 m and embedment lengths from 10 to 49.5 m. The approach employs an interpretable ensemble modeling technique that fuses predictions from diverse individual base learners through Stacked Generalization. The final prediction is generated using a meta-learner to aggregate the predictions of the base learners. Simultaneously, Pareto Multi-Objectives Optimization is utilized to find the optimal 𝑅2 value in both training and testing phases, resulting in the best outcome with the associated parameter set. Consequently, the overarching goal is to improve the accuracy of load–displacement predictions. The model is built upon a substantial dataset encompassing 1178 samples from varied projects across Vietnam, with soil profiles characterized by SPT-N values and static load tests gauging pile head displacement in response to applied loads. To consider the load-bearing complexity, the embedment length is segmented into ten parts, each contributing average SPT indices and corresponding vertical effective stress values to the model. Rigorous validation is achieved via importance-driven sensitivity analysis and partial dependence plots, establishing the model’s adeptness in predicting load–displacement behaviors with precision and reliability. In essence, this proposed soft computing model provides a practical avenue to simulate PHC pile load–displacement behavior, thereby contributing to safer and more operationally efficient structural design in high-rise constructions and beyond.

Files associated with this item cannot be displayed due to copyright restrictions.