This paper addresses the challenge of reducing operation and maintenance costs in wind farms through structural health monitoring. A population-based approach is proposed, implementing a single system for multiple structures (turbines). This system utilizes a Gaussian process-based spatial autoregressive model (GP-SPARX), which captures spatial and temporal correlations between turbines induced by the wake effect. This approach improves the predictability of models and can potentially be applied in a health monitoring system. The study emphasizes the necessity of predictive maintenance strategies for offshore wind farms due to their higher operation and maintenance costs.
Publication date: 16 Oct 2023
Project Page: https://arxiv.org/abs/2310.10555
Paper: https://arxiv.org/pdf/2310.10555
