Spatio-Temporal Forecasting of Retaining Wall Deformation: Mitigating Error Accumulation via Multi-Resolution ConvLSTM Stacking Ensemble

TL;DR

This paper introduces a multi-resolution ConvLSTM ensemble framework that effectively reduces error accumulation and enhances long-term forecasting accuracy of retaining wall deformation during staged excavation, validated through simulations and field data.

Contribution

It presents a novel multi-resolution ConvLSTM stacking ensemble approach that leverages diverse temporal inputs to improve long-horizon geotechnical deformation predictions.

Findings

Ensemble outperforms individual ConvLSTM models in accuracy.

Reduced error propagation in long-term forecasts.

Validated with both numerical simulations and field measurements.

Abstract

This study proposes a multi-resolution Convolutional Long Short-Term Memory (ConvLSTM) ensemble framework that leverages diverse temporal input resolutions to mitigate error accumulation and improve long-horizon forecasting of retaining-structure behavior during staged excavation. An extensive database of lateral wall displacement responses was generated through PLAXIS2D simulations incorporating five-layered soil stratigraphy, two excavation depths (14 and 20 m), and stochastically varied geotechnical and structural parameters, yielding 2,000 time-series deflection profiles. Three ConvLSTM models trained at different input resolutions were integrated using a fully connected neural network meta-learner to construct the ensemble model. Validation using both numerical results and field measurements demonstrated that the ensemble approach consistently outperformed the standalone ConvLSTM models, particularly in long-term multi-step prediction, exhibiting reduced error propagation and improved generalization. These findings underscore the potential of multi-resolution ensemble strategies that jointly exploit diverse temporal input scales to enhance predictive stability and accuracy in AI-driven geotechnical forecasting.