Title:
An Intelligent Approach for Predicting the Performance of Repair Concrete in a Bridge in Northern Ireland from Real-Time Sensor Data
Author(s):
Garg
Publication:
Web Session
Volume:
ws_S23_Garg.pdf
Issue:
Appears on pages(s):
Keywords:
DOI:
Date:
4/2/2023
Abstract:
A section of the Abercorn Bridge in Northern Ireland underwent repairs as a result of corrosion-induced cracking and spalling of the concrete. During the repair, electrical resistance and temperature sensors were installed in the repair concrete to check whether it would protect the steel reinforcement from future corrosion. The sensor system yielded an abundance of complex data, which were difficult to process and arrive at a decision on the future maintenance and/or repair needs. This study used a k-means clustering-based segmented regression approach to extract intelligent features from the real-time resistance and temperature data to obtain the steady-state condition of the resistance. The algorithm determined the optimal number of groups, divided the data into specific clusters, and estimated the best segments for each cluster, while harmonizing against the effects of noise. The results indicate that the repaired concrete was in good condition, thereby created an opportunity for the Northern Ireland Roads Service to make informed decisions based on the best knowledge obtained from the analysis. This systematic approach is feasible and viable to implement for the analysis of data obtained from all types of cementitious materials under any exposure condition and thereby to predict the performance of concrete.