ABSTRACT
The deterioration of concrete structures is influenced by various factors. However, neither the interactions among those factors nor their impacts are explicitly determined. Nowadays, deterioration assessment for concrete structures mainly relies on periodic inspections and on data in maintenance databases that are collected during maintenance. However, superficial analyses of databases are insufficient when maintenance strategies are formulated and/or proper intervention work is implemented. In addition, the factors that affect deterioration show different characteristics and those differences should be taken into account. Usually, the inspection result is indicated by an overall deterioration grade for the concrete structure. However, correlations between potential factors influencing deterioration and the deterioration itself remain unknown. This paper proposes a framework for evaluating the impacts of potential influencing factors on deterioration of concrete structures. A neural network was combined with the Shapley value method to predict deterioration grades, and the factors affecting deterioration were qualitatively and quantitatively calculated. Moreover, the “black box problem” of a neural network was avoided effectively through the adoption of that framework, enabling the uncertainty of these factors to be addressed. In practice, the framework can help to clarify factors that promote or suppress bridge deterioration and can assist in the development of corresponding maintenance strategies.
