Offshore Wind Turbine Jacket Damage Detection via a Siamese Neural Network

Abstract:

This paper states a methodology to detect damage in the support structure of offshore wind turbines. Using vibration-response-only accelerometer measurement, a methodology based on a Siamese convolutional neural network is proposed. In contrast to standard Siamese Neural Networks (SNNs), which are feedforward neural networks, it is proposed to introduce convolutional layers to discerns between healthy and damage structural states. The strategy is validated in an experimental laboratory down-scaled structure. The results demonstrate the feasibility of the proposed methodology.

Año de publicación:

2023

Keywords:

• damage detection
• Data-driven
• Offshore fixed wind turbine
• Jacket structure
• Convolutional neural network
• Damage classification
• Siamese neural network
• Vibration-based SHM

Fuente:

scopus