Structural Health Monitoring (SHM) systems are critical for ensuring the safety and longevity of infrastructure by providing objective evaluations of structural performance and enabling real-time monitoring through measured data. This study aims to develop methods for assessing structural aging based on measured data.
Structural performance degradation due to aging can be identified by observing changes in structural behavior patterns derived from measured responses. These responses, induced by external loads, reflect the structural characteristics. Consequently, pattern changes in structural responses, occurring without significant changes in external factors, indicate changes in structural conditions. This study evaluates structural aging and deterioration through behavioral pattern recognition using a Long Short-Term Memory (LSTM) algorithm.
We applied the LSTM algorithm to learn the structural responses of a suspension bridge model subjected to simulated dynamic loads from multiple vehicles. Trained with time-series data from these simulations, the algorithm evaluated prediction errors to assess aging. These errors, derived from models with incremental performance degradation across multiple regions, indicate changes in structural properties due to aging. The impact of aging varies across different components, affecting the overall system behavior differently. These impacts were quantified and integrated into an overall aging assessment.
Our approach accounted for various factors such as the type of structural response, measurement location, and measurement time, producing a multivariate error dataset representing stages of stiffness degradation. We developed a deep learning algorithm based on this error data to quantify the extent of structural state changes. Testing demonstrated that the algorithm effectively represents the degree of structural performance degradation.
The developed technique shows significant promise for SHM systems to objectively evaluate structural aging. The deep learning algorithm provides a robust framework for quantifying structural performance degradation, offering valuable insights for proactive maintenance strategies. This study presents a novel approach for evaluating structural aging using SHM systems and measured data, contributing to enhanced structural health management and ensuring the safety and longevity of infrastructure.