universidade lusófona

Course on Structural Health Monitoring

Introduce the concept of Structural Health Monitoring (SHM) applied to mechanical and civil structures. The SHM is posed on a statistical pattern recognition paradigm.

Introduce the concept of Structural Health Monitoring (SHM) applied to civil and mechanical and structures. The SHM is posed on a statistical pattern recognition paradigm, where machine learning algorithms are essential to learn (or to model) the structural behavior from the experience (past data), following the same principle of the human brain, in order to analyze Big Data and to perform pattern recognition for damage identification. This concept is rooted in the Artificial Intelligence.

In order to balance the general concept and the applicability of SHM, this course gives some insight about the theory and application of some techniques, such as Gaussian mixture models, kernel principal component analysis, deep learning, regression models, and chaotic systems.

A new learning process is presented based on data generated by both monitoring systems and calibrated finite element models.

A special lecture is given about the application of SHM for bridges in particular and infrastructure in general, as a procedure to support de decision-making process and to reduce the number of structural failures. It is also given an overview on the current bridge management strategies carried out in Portugal, the USA, Angola, and Brazil as well as the main causes of bridges failures around the world. The current challenges and limitations of SHM are highlighted as well as future trends and ongoing research topics.

Outline

  • Introduction to Structural Health Monitoring (SHM)
  • Past, present and future of SHM for bridges and structures in geral
  • Pattern recognition and machine learning for SHM
  • Data- and physics-based approaches
  • Regression models for feature extraction and damage detection
  • Gaussian mixture models for damage detection
  • Principal component analysis-based algorithms for SHM
  • State-space representation for SHM
  • Integration of finite element modeling (FEM) and machine learning for damage detection
  • Challenges, limitations and future trends of SHM

Instructors

  • Eloi Figueiredo (ULHT)
  • Ionut Moldovan (ULHT)
  • Moisés da Silva (UFPA)

This course can be organized as a function of the audience’s background. For more information, please send an e-mail to Eloi Figueiredo.