Hierarchical Federated Learning with Dynamic Clustering and Adaptive Regularization for Robust Infrastructure Inspection

2026-06-01

AI Summary

This work addresses the challenges of data silos caused by privacy constraints in infrastructure visual inspection and the “dual heterogeneity” inherent in federated learning—encompassing both macro-level structural type discrepancies and micro-level non-IID data distributions. To tackle these issues, the authors propose a two-tier collaborative optimization framework for hierarchical federated learning. At the macro level, clients are dynamically clustered based on structural degradation trajectories using a gradient-clustering mechanism that requires no geographic priors. At the micro level, a Dynamic Region-Adaptive Proximal Regularization (DRAPR) module adaptively modulates local updates according to a non-IID intensity score, mitigating catastrophic forgetting of minority damage classes. Experiments on large-scale real-world datasets demonstrate that the proposed method significantly enhances model robustness and diagnostic accuracy, offering the first federated approach capable of simultaneously handling dual heterogeneity while enabling high-precision, privacy-preserving distributed structural damage identification.

Abstract

The deployment of data-driven computer vision models for structural health monitoring (SHM) is heavily constrained by the data silo dilemma due to stringent privacy and security regulations. While federated learning (FL) offers a privacy-preserving collaborative alternative, its application to nationwide infrastructure networks is severely hindered by the challenge of ``double heterogeneity'': macro-level physical divergence across disparate structural types and micro-level statistical imbalances within local datasets. To overcome this challenge, this paper proposes a novel hierarchical federated learning framework. The framework orchestrates a synergistic two-tier optimization strategy. At the macro-level, a dynamic gradient-based clustering mechanism autonomously aggregates distributed clients into specialized expert groups based on their structural degradation trajectories, circumventing the need for prior geographical metadata. Concurrently, at the micro-level, an intra-cluster Dynamic Region-Adaptive Proximal Regularization (DRAPR) module computes a real-time statistical Non-IID Intensity Score for each client. By adaptively modulating a proximal penalty based on local label skewness and gradient divergence, DRAPR effectively calibrates local updates, mitigates client drift, and prevents the catastrophic forgetting of minority damage classes. Comprehensive evaluations on a large-scale, real-world structural inspection dataset demonstrate that the hierarchical integration of macro-clustering and micro-regularization successfully neutralizes dual-level heterogeneity, yielding highly robust and specialized diagnostic models for complex infrastructure inspection.

Problem

• federated learning
• structural health monitoring
• data heterogeneity
• data silo
• infrastructure inspection

Innovation

• Hierarchical Federated Learning
• Dynamic Clustering
• Adaptive Regularization
• Non-IID Intensity Score
• Structural Health Monitoring