TY - GEN
T1 - Regression model for structural health monitoring of a lab scaled bridge
AU - Hassan, Rubayet
AU - Athar, Seyyed Pooya Hekmati
AU - Taheri, Mohammad
AU - Cesmeci, Sevki
AU - Taheri, Hossein
N1 - Publisher Copyright:
© 2021 SPIE.
PY - 2021
Y1 - 2021
N2 - The interest in observation of the dynamic behavior of bridges have been increasing in the recent years. The movement of bridge deck plays a significant role in the safety of bridges. In this project work, a direct and indirect sensor mounted on the bridge structure and on the passing vehicle are used for structural health monitoring. The overall study has been implemented based on six reliable approaches, including Gradient Boosting regression, Random Forest Regression, Ridge Regression, Support Vector Regression, Elastic Net Regression, XGBoost Regression and Support Vector Regression to get accurate results of prediction for structural health condition. For each of these regression models, the following performance evaluations are obtained: Mean Square Error (MSE), Root Mean Square Error (RMSE) and R-squared. After obtaining all performance evaluations, the comparison of each of these metrics are done for all the six regressors. Finally, by using a Voting Regression, these six regression models are combined and used to train the entire dataset and predict on the test set. By using voting regression an ensemble model is proposed for this experiment.
AB - The interest in observation of the dynamic behavior of bridges have been increasing in the recent years. The movement of bridge deck plays a significant role in the safety of bridges. In this project work, a direct and indirect sensor mounted on the bridge structure and on the passing vehicle are used for structural health monitoring. The overall study has been implemented based on six reliable approaches, including Gradient Boosting regression, Random Forest Regression, Ridge Regression, Support Vector Regression, Elastic Net Regression, XGBoost Regression and Support Vector Regression to get accurate results of prediction for structural health condition. For each of these regression models, the following performance evaluations are obtained: Mean Square Error (MSE), Root Mean Square Error (RMSE) and R-squared. After obtaining all performance evaluations, the comparison of each of these metrics are done for all the six regressors. Finally, by using a Voting Regression, these six regression models are combined and used to train the entire dataset and predict on the test set. By using voting regression an ensemble model is proposed for this experiment.
KW - Bridge
KW - Combined Direct-Indirect Method
KW - Ensemble
KW - Mean Square Error (MSE)
KW - R-squared
KW - Regression
KW - Root Mean Square Error (RMSE)
KW - Structural Health Monitoring (SHM)
KW - Voting
UR - http://www.scopus.com/inward/record.url?scp=85107184560&partnerID=8YFLogxK
U2 - 10.1117/12.2592037
DO - 10.1117/12.2592037
M3 - Conference article
AN - SCOPUS:85107184560
T3 - Proceedings of SPIE - The International Society for Optical Engineering
BT - NDE 4.0 and Smart Structures for Industry, Smart Cities, Communication, and Energy
A2 - Meyendorf, Norbert G.
A2 - Farhangdoust, Saman
A2 - Niezrecki, Christopher
PB - SPIE
T2 - NDE 4.0 and Smart Structures for Industry, Smart Cities, Communication, and Energy 2021
Y2 - 22 March 2021 through 26 March 2021
ER -