What does this research mean for the field?

The Gated Recurrent Unit (GRU) model optimized with the whale optimization algorithm (GRU-WOA) provides highly accurate predictions of diaphragm-wall settlement in deep excavations, outperforming other standard machine learning models. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

To evaluate multiple predictive models and metaheuristic optimization algorithms for forecasting diaphragm-wall settlement and screening relative deformation anomalies during deep excavation.

August 30, 2026BuildingsOpen Access

Diaphragm-Wall Settlement Prediction and Relative Anomaly Screening for Deep Excavations Using Multi-Model Comparison and Intelligent Optimization

View Full Paper

Ask AI

Bookmark

Authors

YXYuhang Xu XAXinying AiDongguan University of Technology JFJian Fang

Discussion

Member takes

Implication

Comparative modeling study reveals superior diaphragm-wall settlement prediction using optimized gated recurrent units in deep excavations, indicating effective support for targeted monitoring.

Key Points

To evaluate multiple predictive models and metaheuristic optimization algorithms for forecasting diaphragm-wall settlement and screening relative deformation anomalies during deep excavation.
Analyzed cumulative vertical settlement across 23 diaphragm-wall monitoring points from the Tianjin Goldin Finance 117 deep excavation project.
Evaluated seven baseline models—a naïve persistence baseline, ARIMA, KNN, MLP, GRU, Transformer, and XGBoost—using a five-fold rolling-origin expanding-window validation scheme.
Optimized the top-performing baseline model using the crow search algorithm (CSA), genetic algorithm (GA), and whale optimization algorithm (WOA).
Baseline GRU achieved the best initial predictive accuracy, with a mean R² of 0.9172, MAE of 0.1009 mm, RMSE of 0.1341 mm, and MAPE of 0.5875%.
Whale optimization algorithm coupling (GRU-WOA) produced the highest performance among all configurations, reaching a mean R² of 0.9289 and an RMSE of 0.1215 mm.
Relative anomaly levels derived from predicted settlement-change rates successfully mapped the temporal concentration and spatial clustering of ground deformation.

Cite This Study

Xu et al. (2026) studied this question.

synapsesocial.com/papers/6a93f0e06c1a8fb52e79d5bc https://doi.org/10.3390/buildings16173441

View Full Paper

Ask AI

Bookmark

Implication

Comparative modeling study reveals superior diaphragm-wall settlement prediction using optimized gated recurrent units in deep excavations, indicating effective support for targeted monitoring.

Diaphragm-Wall Settlement Prediction and Relative Anomaly Screening for Deep Excavations Using Multi-Model Comparison and Intelligent Optimization

Discussion

Member takes

Implication

Key Points

Cite This Study

Diaphragm-Wall Settlement Prediction and Relative Anomaly Screening for Deep Excavations Using Multi-Model Comparison and Intelligent Optimization

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider