A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
AbstractA bridge-deterioration approach is to predict the condition ratings and the deterioration pattern of bridge elements for determining optimal maintenance strategies and estimating future funding requirements. To effectively predict long-term bridge performance, an advanced integrated deterioration approach has been developed that incorporates a time-based model, a state-based model with the Elman neural network (ENN) and a backward prediction model (BPM). The proposed approach involves the categorization of the selected inspection records by bridge components, material types, traffic volume, and the construction era. The primary advantage of such categorization is to group similar components together, thereby identifying the common deterioration patterns. A selection process embedded in the proposed approach offers the ability to automatically select the most appropriate model for predicting future bridge condition ratings. To demonstrate the advantage of the proposed approach in predicting long-term bridge performances, case studies are performed using available inspection records. To compare the performance of the proposed approach against the standard Markovian-based deterioration procedure in predicting future bridge condition ratings, a total of 40 bridges with 464 bridge substructure inspection records are selected and used as input. The predicted outcomes are validated by a cross-validation process, which demonstrates that the proposed approach is more accurate than the standard Markovian-based procedure.
AbstractA bridge-deterioration approach is to predict the condition ratings and the deterioration pattern of bridge elements for determining optimal maintenance strategies and estimating future funding requirements. To effectively predict long-term bridge performance, an advanced integrated deterioration approach has been developed that incorporates a time-based model, a state-based model with the Elman neural network (ENN) and a backward prediction model (BPM). The proposed approach involves the categorization of the selected inspection records by bridge components, material types, traffic volume, and the construction era. The primary advantage of such categorization is to group similar components together, thereby identifying the common deterioration patterns. A selection process embedded in the proposed approach offers the ability to automatically select the most appropriate model for predicting future bridge condition ratings. To demonstrate the advantage of the proposed approach in predicting long-term bridge performances, case studies are performed using available inspection records. To compare the performance of the proposed approach against the standard Markovian-based deterioration procedure in predicting future bridge condition ratings, a total of 40 bridges with 464 bridge substructure inspection records are selected and used as input. The predicted outcomes are validated by a cross-validation process, which demonstrates that the proposed approach is more accurate than the standard Markovian-based procedure.
Prediction of Long-Term Bridge Performance: Integrated Deterioration Approach with Case Studies
Bu, G. P (author) / Guan, H / Lee, J. H / Blumenstein, M / Loo, Y. C
2015
Article (Journal)
English
Prediction of Long-Term Bridge Performance: Integrated Deterioration Approach with Case Studies
| British Library Online Contents | 2015
Prediction of Long-Term Bridge Performance: Integrated Deterioration Approach with Case Studies
| BASE | 2015
| British Library Conference Proceedings | 2012