Statistical and Clustering Analysis for Disturbances: A Case Study of Voltage Dips in Wind Farms

• Published in: IEEE transactions on power delivery Vol. 31; no. 6; pp. 2530 - 2537
• Main Authors: Garcia-Sanchez, T., Gomez-Lazaro, E., Muljadi, E., Kessler, M., Molina-Garcia, A.
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.12.2016; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Circuit faults; Clustering methods; Dipping; Disturbances; Electric potential; Evolution; Graphical representations; Power quality; POWER TRANSMISSION AND DISTRIBUTION; Principal component analysis; Principal components analysis; Trajectories; Trajectory; Voltage; voltage dip; Voltage fluctuations; Voltage measurement; Wind farms; Wind power
• ISSN: 0885-8977; 1937-4208; 1937-4208
• DOI: 10.1109/TPWRD.2016.2522946

This paper proposes and evaluates an alternative statistical methodology to analyze a large number of voltage dips. For a given voltage dip, a set of lengths is first identified to characterize the root mean square (rms) voltage evolution along the disturbance, deduced from partial linearized time intervals and trajectories. Principal component analysis and K-means clustering processes are then applied to identify rms-voltage patterns and propose a reduced number of representative rms-voltage profiles from the linearized trajectories. This reduced group of averaged rms-voltage profiles enables the representation of a large amount of disturbances, which offers a visual and graphical representation of their evolution along the events, aspects that were not previously considered in other contributions. The complete process is evaluated on real voltage dips collected in intense field-measurement campaigns carried out in a wind farm in Spain among different years. The results are included in this paper.