Using Graph Neural Networks and Frequency Domain Data for Automated Operational Modal Analysis of Populations of Structures

TL;DR

This paper introduces a GNN-based approach for automated modal analysis of structures using frequency domain data, enabling transfer learning across similar structures with improved speed and accuracy over traditional methods.

Contribution

The work presents a novel GNN model for structural modal identification that leverages population-based transfer learning and frequency domain data, outperforming classic methods.

Findings

GNN accurately identifies modal properties of unseen structures within the same population.

The proposed method outperforms Frequency Domain Decomposition in speed.

The GNN model maintains good accuracy even with noisy and sparse data.

Abstract

The Population-Based Structural Health Monitoring (PBSHM) paradigm has recently emerged as a promising approach to enhance data-driven assessment of engineering structures by facilitating transfer learning between structures with some degree of similarity. In this work, we apply this concept to the automated modal identification of structural systems. We introduce a Graph Neural Network (GNN)-based deep learning scheme to identify modal properties, including natural frequencies, damping ratios, and mode shapes of engineering structures based on the Power Spectral Density (PSD) of spatially-sparse vibration measurements. Systematic numerical experiments are conducted to evaluate the proposed model, employing two distinct truss populations that possess similar topological characteristics but varying geometric (size, shape) and material (stiffness) properties. The results demonstrate that, once trained, the proposed GNN-based model can identify modal properties of unseen structures within the same structural population with good efficiency and acceptable accuracy, even in the presence of measurement noise and sparse measurement locations. The GNN-based model exhibits advantages over the classic Frequency Domain Decomposition (FDD) method in terms of identification speed, as well as against an alternate Multi-Layer Perceptron (MLP) architecture in terms of identification accuracy, rendering this a promising tool for PBSHM purposes.