Kernel-based Column Drift Ratios Prediction in Highway Bridges

Document Type

Conference Proceeding

Publication Date

1-1-2024

Abstract

This study focuses on quantifying the critical parameter of column drift ratio in bridge engineering and proposes a novel kernel-based regression approach to enhance the performance-based seismic assessment of bridge systems. Traditionally, analytical methods in this field have relied on power-law functions of a single ground motion intensity measure. However, recent research has explored alternative models, though the application of machine learning (ML) approaches for bridge demand quantification and performance-based seismic assessment remains largely untapped. To address this gap, we introduce an advanced ML algorithm, specifically a kernel-based Gaussian regression approach, to estimate the column drift ratio metric for bridges. The effectiveness of the proposed model is demonstrated through its application to a representative class of highway bridges in California. The results reveal that the kernel-based model performs comparably to conventional approaches, underscoring its significance in efficiently estimating column drift ratio within the performance-based engineering framework. Importantly, the model's implications extend beyond accurate estimation, as it can inform infrastructure resilience assessments and facilitate rapid decision-making processes post-seismic events. By harnessing the capabilities of ML algorithms, this approach presents a compelling alternative to conventional methods, advancing earthquake engineering practices and providing valuable insights into the behavior of bridge systems under seismic conditions.

Publication Source (Journal or Book title)

Proceedings of the World Congress on New Technologies

This document is currently not available here.

Share

COinS