Forecasting bridge deck conditions using fuzzy regression analysis

A good bridge management system requires an accurate and efficient bridge performance and prediction model. Ordinary regression models have been extensively used to forecast future infrastructure conditions. However, data on current bridge conditions obtained from inspectors are inherently subjective and non-crisp; thus ordinary regression techniques are incapable of predicting future bridge performance when inspection data are not numerical. This paper presents a multiple fuzzy linear regression model for the estimation of bridge deterioration conditions. The proposed model can effectively tackle non-crisp data and a mixture of fuzzy data and crisp data. An empirical case study using bridge inspection data from Taiwan is used to examine the variables contributing to deterioration of concrete decks. The results demonstrate the capability and effectiveness of the proposed model, which can assist bridge managers to better predict bridge deck performance.