Spatio-Temporal Forecasting of Retaining Wall Deformation: Mitigating Error Accumulation via Multi-Resolution ConvLSTM Stacking Ensemble
arXiv:2603.10453v2 Announce Type: replace 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 2,00
The paper demonstrates an ongoing trend in applying advanced AI techniques like multi-resolution ConvLSTM ensembles to specific, complex engineering challenges that were previously difficult to model accurately.
This development improves predictive capabilities for critical infrastructure, such as retaining walls, enhancing safety and potentially reducing costs associated with monitoring and maintenance in civil engineering.
The ability to accurately forecast long-horizon deformation of structures using AI mitigates error accumulation, impacting asset management and risk assessment in large-scale construction and existing infrastructure.
- · Civil Engineering Consultancies
- · Construction Firms
- · Infrastructure Monitoring Companies
- · AI/ML Engineering Firms
- · Traditional Geotechnical Modeling Services (less accurate)
More reliable predictions for the stability of retaining structures in complex environments.
Reduced incidence of structural failures and associated economic losses and human risks in construction and urban development.
Potential for broader application of similar multi-resolution AI ensemble methods to other complex physical forecasting problems across various engineering disciplines.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG