2602.07031v1 Feb 03, 2026 cs.LG

지연된 역호환 물리 기반 신경망을 활용한 불포화 토양 침하 분석

Lagged backward-compatible physics-informed neural networks for unsaturated soil consolidation analysis

Lanzhou
Lanzhou
Citations: 478
h-index: 12
Dong Li
Dong Li
Citations: 55
h-index: 3
Shuai Huang
Shuai Huang
Citations: 17
h-index: 3
Yapeng Cao
Yapeng Cao
Citations: 19
h-index: 1
Yujun Cui
Yujun Cui
Citations: 1
h-index: 1
Xiaobin Wei
Xiaobin Wei
Citations: 37
h-index: 2
Hongtao Cao Department of Civil
Hongtao Cao Department of Civil
Citations: 0
h-index: 0
Environmental
Environmental
Citations: 0
h-index: 0
Infrastructure Engineering
Infrastructure Engineering
Citations: 1
h-index: 1
George Mason University
George Mason University
Citations: 793
h-index: 16
Fairfax
Fairfax
Citations: 127
h-index: 5
Va
Va
Citations: 17
h-index: 1
Usa National Institute of Natural Hazards
Usa National Institute of Natural Hazards
Citations: 0
h-index: 0
M. O. E. Management
M. O. E. Management
Citations: 0
h-index: 0
Beijing
Beijing
Citations: 202
h-index: 6
C. C. O. Science
C. C. O. Science
Citations: 147
h-index: 5
Frozen Soil Engineering
Frozen Soil Engineering
Citations: 0
h-index: 0
Northwest Institute of Eco-Environment
Northwest Institute of Eco-Environment
Citations: 0
h-index: 0
Resources
Resources
Citations: 0
h-index: 0
Chinese Academy of Sciences
Chinese Academy of Sciences
Citations: 6,190
h-index: 34
China
China
Citations: 161
h-index: 7
N. Laboratory
N. Laboratory
Citations: 13
h-index: 1
'Ecole Nationale des Ponts et Chauss'ees
'Ecole Nationale des Ponts et Chauss'ees
Citations: 0
h-index: 0
2. Marne-la-Vall'eeCedex
2. Marne-la-Vall'eeCedex
Citations: 0
h-index: 0
France Navier Laboratory
France Navier Laboratory
Citations: 0
h-index: 0
France School of Civil Engineering
France School of Civil Engineering
Citations: 0
h-index: 0
Hebei University of Engineering
Hebei University of Engineering
Citations: 0
h-index: 0
Handan
Handan
Citations: 5
h-index: 1
China Department of Nuclear Engineering
China Department of Nuclear Engineering
Citations: 98
h-index: 2
Zhejiang University of Technology
Zhejiang University of Technology
Citations: 4
h-index: 1
Hangzhou
Hangzhou
Citations: 16
h-index: 3

본 연구는 장기 하중 조건 하에서 일차원 불포화 토양의 침하 현상을 시뮬레이션하고 역추적하기 위한 지연된 역호환 물리 기반 신경망(LBC-PINN)을 개발합니다. 다중 스케일 시간 영역에서 공기 및 수압 소산의 복잡성을 해결하기 위해, 본 프레임워크는 로그 스케일 시간 분할, 지연된 호환성 손실 강제, 그리고 세그먼트별 전이 학습을 통합합니다. 전방 분석에서, 권장되는 분할 방식을 사용한 LBC-PINN은 공극 기압 및 공극 수압의 변화를 정확하게 예측합니다. 모델 예측 결과는 유한 요소법(FEM) 결과와 비교하여 검증되었으며, 최대 10^10초까지의 시간 구간에서 평균 절대 오차가 10^-2 이하로 나타났습니다. 특성 공기 소산 시간을 기반으로 한 단순화된 분할 전략은 예측 정확도를 유지하면서 계산 효율성을 향상시킵니다. 민감도 분석 결과, 공기 투과율과 수 투과율 비율이 10^-3에서 10^3까지 변하는 경우에도 본 프레임워크의 안정성이 확인되었습니다.

Original Abstract

This study develops a Lagged Backward-Compatible Physics-Informed Neural Network (LBC-PINN) for simulating and inverting one-dimensional unsaturated soil consolidation under long-term loading. To address the challenges of coupled air and water pressure dissipation across multi-scale time domains, the framework integrates logarithmic time segmentation, lagged compatibility loss enforcement, and segment-wise transfer learning. In forward analysis, the LBC-PINN with recommended segmentation schemes accurately predicts pore air and pore water pressure evolution. Model predictions are validated against finite element method (FEM) results, with mean absolute errors below 1e-2 for time durations up to 1e10 seconds. A simplified segmentation strategy based on the characteristic air-phase dissipation time improves computational efficiency while preserving predictive accuracy. Sensitivity analyses confirm the robustness of the framework across air-to-water permeability ratios ranging from 1e-3 to 1e3.

0 Citations
0 Influential
17 Altmetric
85.0 Score

No Analysis Report Yet

This paper hasn't been analyzed by Gemini yet.

댓글

댓글을 작성하려면 로그인하세요.

아직 댓글이 없습니다. 첫 번째 댓글을 남겨보세요!