Wednesday, July 14th, 2021. Time: 12 noon

Online session: https://meet.google.com/qjo-sttx-dgo

See video on Youtube

ABSTRACT

Machine Learning (ML) is an emerging technology with applications in multiple fields. The concepts and techniques developed in the community of computer science is now in use by an increasing set of disciplines. The common feature of all applications of ML is the need for data for model fitting. These data may come from sensors, but they can also be generated with numerical models.

The benefits of these tools attracted the attention of the numerical modelling community. As an example, 9 minisimposia were specifically focused on ML in the last 14th World Congress in Computational Mechanics. However, there is still a lack of communication between both disciplines: there is some confusion regarding the terms used, the possibilities of the technology and its true potential. A similar situation exists in practical civil engineering: the approach is becoming popular, but many practitioners are still reluctant to modify their traditional procedures.

In this seminar, some concepts related to machine learning will be discussed from a practical engineering viewpoint. In addition, examples of application in different settings will be presented, both alone and in combination with numerical models. They include:

• generation of prediction models for dam behavior modelling based on monitoring data;
• identification of behavior patterns in dam response with classification models;
• analysis of the seismic response of gravity dams with heterogeneous concrete;
• estimation of the discharge capacity of arched labyrinth spillways.
• calibration of constitutive laws in DEM models.

SPEAKER CV

MSc Civil Engineering (UPM, 2002) and PhD on Structural Analysis (UPC, 2017). Associate Research Professor at CIMNE and head of the group on “Machine Learning in Civil Engineering”. He has a background in engineering consultancy, focused on hydraulic engineering and dam safety. Since he joined CIMNE in 2009, his research lines involved the combination of data-driven and numerical modelling to solve problems in different fields. In his PhD Thesis, he developed a prototype application for dam safety assessment based on machine learning which obtained the innovation award promoted by Verbund Hydropower. He is member of the Technical Committees on “Numerical Modelling” and “Dam Surveillance” of the Spanish Committee on Large Dams (SPANCOLD). He is author of 20 papers in indexed journals and was IP in 25 research projects in cooperation with industrial partners.

See Seminar Programme 2021