Predicting bridge elements deterioration, using Collaborative Gaussian Process Regression

Roadway and railway bridges are not only integral, but also vulnerable parts of terrestrial transport networks. Structural failures of bridges may lead to disastrous consequences on users and society at large. Bridge predictive deterioration models are extremely important for effective maintenance decision-making. However, the lack of enough inspection data between maintenance activities of a bridge complicates the development of accurate predictive models. Presented herein is a Gaussian Process Regression (GPR) based collaborative model for predicting the condition of bridge elements with limited available inspection data per bridge. This model has been applied in 137 bridge decks, showing that collaborative prognosis has the potential to predict the condition of different types of bridge elements, composing different types of bridges