Wind turbine performance monitoring based on hybrid clustering method

Abstract

Due to the largely increasing demand for electrical power, other sources of energy have to be sought and wind power is one of those. Wind farms from around the world have continued to thrive due to its cost-effectiveness and benefits. However, an utmost concern for wind farm operators is to keep the turbines in good working conditions in order to produce power at the most optimal level. For wind turbines, a maintenance activity can be very costly; therefore, it should be carried out from a well-guided decision. An accurate monitoring of a turbine's performance is instrumental for detecting a potentially deteriorating state. In this paper, we present a performance monitoring system for wind turbines based on ANFIS, a hybrid neuro-fuzzy system. By taking advantage of the combined strengths of neural networks and fuzzy inference systems, an accurate modeling of wind turbine performance is established. Its performance is evaluated using actual SCADA and it proves to be a favorable alternative to conventional modeling techniques. © 2013 Springer Science+Business Media Dordrecht.