Laboratory testing campaigns for the wind energy industry play an essential role in testing innovative control strategies and digital twin applications. But incidents during testing can be detrimental and might cause project delays and damage to expensive equipment. We propose an anomaly detection scheme for laboratory experiments that are developed and tested to enhance reaction time and prediction quality, reducing the likelihood of damage to equipment due to human error or software malfunction.

This study utilizes a method to analyze power conversion dynamics across different operational states, addressing non-stationarity with a correlation matrix algorithm. Findings reveal distinct dynamics for each state, emphasizing their impact on system behavior and offering a solution for hysteresis effects in power conversion dynamics.