The operation and maintenance of wind turbines benefit from reliable information on the wind turbine condition. Data-driven models use data from the supervisory data acquisition system. In particular, great performance is reported for artificial intelligence models. However, the lack of interpretability limits their effective industrial implementation. The present work introduces a new condition-monitoring approach for wind turbines featuring a built-in visualization tool that confers interpretability upon the model outcomes. The proposed approach is based on a supervised implementation of the variational autoencoder model, which allows the projection of the wind turbine system onto a low-dimensional representation space. Three outcomes follow from such representation: a health indicator for the early detection of abnormal conditions, a classifier providing the diagnosis status, and a visualization tool depicting the wind turbine condition as a trajectory in a 2D plot. The approach is implemented with a vast database. Two case studies demonstrate the potential of the proposed approach. The proposed health indicator detects the main bearing overtemperature 11 days before the control system alarm, one week earlier than a competing approach. Study cases illustrate that the built-in visualization tool enhances the interpretability and trust in the model outcomes, thus supporting wind turbine operation and maintenance.

The rapidly increasing installed capacity of Wind Turbines (WTs) worldwide emphasizes the need for Operation and Maintenance (O&M) strategies favoring high availability, reliability, and cost-effective operation. Optimal decision-making and planning are supported by WT health condition analyses based on data from the Supervisory Control and Data Acquisition (SCADA) system. However, SCADA data are highly imbalanced, with a predominance of healthy condition samples. Although this imbalance can negatively impact analyses such as detection, Condition Monitoring (CM), diagnosis, and prognosis, it is often overlooked in the literature. This review specifically addresses the problem of SCADA data imbalance, focusing on strategies to mitigate this condition. Five categories of such strategies were identified: Normal Behavior Models (NBMs), data-level strategies, algorithm-level strategies, cost-sensitive learning, and data augmentation techniques. This review evidenced that the choice among these strategies is mainly dictated by the availability of data and the intended analysis. Moreover, algorithm-level strategies are predominant in analyzing SCADA data because these strategies do not require the costly and time-consuming task of data labeling. An extensive public SCADA database could ease the problem of abnormal data scarcity and help handle the problem of data imbalance. However, long-dated requests to create such a database are still unaddressed.

This work introduces a new condition monitoring approach for complex systems based on a standardized latent space representation. Latent variable models such as the variational autoencoders are widely used to analyze systems described by a high-dimensional physical space. The encoding of such space defines a low-dimensional and physically representative latent space. Of note, however, the latent space obtained for complex systems operating under multiple conditions is often difficult to exploit in defining an efficient Health Index, thanks to the non-deterministic and hyperparameter-dependent nature of the latent space. In addition, the distribution of the healthy cluster is not known a priori. The original contribution of this paper is to use the Nataf isoprobabilistic transform to map the latent space into a standardized space. This normalizes the spatial structure of the latent space and relaxes the model’s sensitivity to hyperparameters during the learning process. Moreover, the characterization of the healthy condition in the standard Nataf space leads to the definition of two complementary health indices suitable for complex systems. An implementation in two case studies demonstrates the potential of the proposed approach. First, the approach was successfully applied within NASA’s Commercial Modular Aero-Propulsion System Simulation dataset. The second case study consisted in analyzing multiple degradation in operating wind turbines. Encouraging results emerge from both case studies, with critical conditions being detected significantly earlier than in competing approaches. The proposed approach can be generalized to complex systems equipped with multiple sensors, and overcomes difficulties related to latent space analysis of multiple condition systems.

This study introduces a classification-based prognostic framework to anticipate temperature threshold exceedance events that lead to automatic shutdowns in wind turbines. The framework assesses whether an observed temperature anomaly is likely to evolve into such an event and relies exclusively on Supervisory Control and Data Acquisition data combined with Deep Learning and Conformal Prediction for uncertainty quantification. To characterize the temporal evolution of the exceedance tendency, we propose a novel metric, the Instantaneous Event Intensity, derived from confidence intervals produced by Conformal Prediction. By aggregating Instantaneous Event Intensity values over time, we define the Cumulative Event Intensity, which integrates the event magnitude. The Cumulative Event Intensity is then compared with a predefined event threshold to support binary decision-making on whether the component will experience a temperature threshold exceedance event. The proposed approach is validated on 207 wind turbines from 20 wind farms, covering three manufacturers and three distinct components, detecting 87.25% of temperature threshold exceedance events 15 days in advance, with a false positive rate below 21%. Notably, several events can be anticipated up to three months earlier, highlighting the robustness of the early-warning capability.