A nonlinear regression-based machine learning model for predicting concrete bridge deck condition

Abstract

Understanding the process of concrete bridge deck deterioration and evaluating its condition are important for maintaining a healthy transportation infrastructure and for allocating the necessary funds for bridge maintenance, rehabilitation, or reconstruction actions. Therefore, it is important to investigate the factors impacting bridge condition to enable the development of predictive techniques. The main objective of this paper is to study the impact of average daily traffic (ADT), age, and deck area on the concrete bridge deck deterioration. Michigan concrete bridge deck condition data for the past 25 years was analyzed to determine the impact of these factors on concrete decks. An optimum machine learning algorithm that is based on nonlinear regression modeling has been developed to predict the deterioration rates of bridge decks under these impacting factors. This study has revealed that ADT, age, and deck area have a significant effect on the deterioration of concrete bridge decks.