Strategies for model updating and structural health monitoring of wind turbine blades

December 13, 2022

One of the most appealing renewable energy power sources is wind. The greatest challenge of this growing sector is monitoring the condition of wind turbine structures. They are vulnerable to damage and deterioration, because they operate under large mechanical and aerodynamic loads and extreme environmental conditions. Developing Structural Health Monitoring (SHM) strategies is crucial to ensure that damages are detected effectively. In this thesis, three Machine Learning (ML) damage detection methodologies are tested: Multivariate Gaussian Anomaly Detection (MGAD), Principal Components Analysis (PCA) and Anomaly Detection Autoencoder (ADAE). These techniques were implemented to recognize deviating patterns from the healthy state to the damaged state of a structure. The data were acquired experimentally from a Glass Fiber Reinforced Polymer (GFRP) scaled blade and features were extracted, such as modal parameters, Frequency Response Funtion (FRF) and acceleration time signals. In response to the data scarcity barrier imposed by the experimental data on the potential of ML algorithms, the second part of this thesis focuses on the Finite Element Method (FEM). For the use of simulation data to be successfully applied to real situations, it needs to be a reliable representation of reality. One way of accomplishing this is by developing model updating strategies. Making use of a Finite Element (FE) model of the blade studied before, its parameters are tuned in order to reduce the differences between the experimental response data and the FE model.

Authors in the community:

• 

  I

  Inês Oliveira Ribas Fernandes

  ist189673

Supervisors of this institution:

• 

  N

  Nuno Silvestre

  ist13506

mechanical-engineering - Mechanical engineering

eng - English

Instituto Superior Técnico