Deep convolutional neural network for the analysis of bridge element data

Bridge management engineers in the United States started employing bridge elements conditions for improving the assessment of bridge assets. However, the analysis of bridge element conditions requires a considerable amount of data collected over a relatively long period of time. Such data are still limited when compared to bridge condition ratings taken from the National Bridge Inventory (NBI), which have been assembled for decades. There is a need to correlate element condition data to NBI ratings to help establishing trends for more reliable predictions of deterioration rates. The objective of this paper is to perform the joint analysis of bridge element condition data and NBI ratings to back-map NBI deterioration curves into element deterioration profiles using deep convolutional neural networks. The proposed approach improves the accuracy of current methods used to convert element conditions to NBI ratings by almost 30% providing more reliable estimates of bridge element deterioration rates using NBI data. Results of case studies from the State of NJ as well as from regions in the Northeast of USA are presented.