Comparative Study of the Dynamic Back-Analysis Methods of Concrete Gravity Dams Based on Multivariate Machine Learning Models

ABSTRACT

Two different back-analysis frameworks based on multivariate machine learning models used to determine the material dynamic parameters of concrete gravity dams are proposed. For the framework I, the back-analysis is performed by solving an optimization problem and a multivariate machine learning model is trained to replace the FEM calculation during the optimization process. While the framework II uses a multivariate machine learning model directly and the material dynamic parameters are predicted using the machine learning mode. By using a numerical example and an experimental investigation, the robustness, accuracy, computation efficiency of these proposed back-analysis methods is verified.

KEYWORDS:

• Concrete Gravity Dams
• Back-Analysis
• Machine Learning
• Multiple-Outputs Support Vector Machines
• Multivariate Relevant Vector Machines

Acknowledgments

The authors are grateful to Fernando Pérez-Cruz and Youngmin Ha for making their MATLAB implementations of M-SVM and MRVM freely available, respectively.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China (Grant Nos. 51409205), the Project Funded by China Postdoctoral Science Foundation (Grant No. 2015M572656XB), the Open Foundation of State Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering (Grant No. 2014491011), and the innovative research team of institute of water resources and hydro-electric engineering, Xi’an University of Technology (Grant No. 2016ZZKT-14).