Application of Bayesian Network for concrete bridge deck condition rating

One of the important efforts in bridge management systems is to find current and/or future condition of the bridge deck based on the inspection results. Bayesian Networks are able to illustrate interrelationship of uncertain information propagation for almost any real world phenomena. Bayesian Networks are robust to missing data therefore they combine existing information well. Although each variable or input data has only a limited chance of giving a correct interpretation, the combination of all the inputs' chances, usually increases the probability of a valid interpretation. Adding any new decision variable to the relationships of a Bayesian Network is practically possible whenever its necessity becomes apparent. In this paper it is shown that probabilistic representation of visual inspection results may be useful in construction of the Bayesian Network for concrete bridge deck condition rating. The proposed method can be a practical tool for decision makers with incomplete and imprecise data.