Anomaly Detection in Wind Turbines: Analysis of Data Imbalance and Temporal Effects

Abstract: In engineering and systems monitoring, an anomaly is defined as a rare event. Detecting these anomalies, which deviate significantly from expected or normal system behavior, is crucial for identifying issues in various contexts, including computer systems and industrial processes.

This project focuses on detecting anomalies in the operation of wind turbines used for renewable energy generation. Environmental and operational factors can affect turbine performance, leading to occasional anomalies. Early detection is vital for ensuring efficiency and safety, enabling predictive maintenance, optimizing performance, and extending turbine lifespan.

Detecting anomalies in real-world data is challenging due to the rarity and unpredictability of such events. Machine learning, specifically supervised learning, is commonly used for this task. In our study, each data point is labeled as anomalous or normal by experts, ensuring reliable results.

We address two main challenges: handling class imbalance in the dataset, which is typical for this type of application, and analyzing the effectiveness of treating data as time series to improve anomaly detection accuracy.