Investigating power loss in a wind turbine using real-time vibration signature

Abstract

Wind energy is one of the fastest-growing sources of renewable energy; hence, a comprehensive understanding of all the environmental and mechanical factors that influence the power output from wind turbines is essential. Despite vibration being a crucial degradation indicator (and thus linked to wind turbine health), its role in influencing power loss is rarely explored. In this paper, we propose to use a novel feature, the vibration index, derived from the vibration data, to understand the power loss in wind turbines. The methodology adopted in this work involves modelling the gearbox health as the loss of generator speed using a data-driven methodology with vibration index as input. It was observed that vibration had a significant impact on the generator speed. Vibration of different parts of the turbine hub was found to lead to a loss of rotational speed in the generator, which in turn led to power loss. Wind turbine vibration was found to explain about 57% (R2: 0.57) of the loss in speed. Further, the generator speed is linearly mapped to power with an accuracy of 85% (R2:0.85), and consecutively, the predicted loss in the generator speed is mapped to the power loss in the turbine. The accuracy of power loss prediction was about 55% (R2: 0.55). The proposed methodology is applied to data obtained from a real-world onshore wind farm located in the state of Gujarat (India). This approach offers a data-driven and efficient methodology that augments the understanding of wind turbine operations and the complex relationship between vibration and the power output of a wind turbine. The outcomes of this research can contribute to improving wind turbine performance and enhancing decision-making for system maintenance and operation.