SCADA data based realistic simulation framework to evaluate environmental impact on performance of wind turbine condition monitoring systems
Résumé
Wind turbines are an integral part of renewable energy based power generation targets all over the world. With expanding fleets, varying geographical locations, improving technologies, competitive markets and a variety of manufactures to choose from, wind farm operators often end up with heterogeneous fleets. This paper presents a methodology to comprehensively evaluate the performance of fault indicators for wind turbines condition monitoring using power curves. A comprehensive analysis is done by taking into account various factors that can influence performance indicators such as the differences between the geographical locations of the wind farms and multiple fault signatures. A controlled and realistic simulation framework is presented and used to create a database covering multiple wind farms from different geographical conditions having distinct environmental profiles. Realistic fault signatures are used for condition monitoring and performance analysis using two fault detection approaches. Results show a trend in performance indicators associated to the distinct environmental profiles of the wind farms under observation. The results also highlight the limitations of local and site specific detection and performance analysis approaches and identify the distinct environmental profiles as key phenomenon for consideration in performance evaluation of fault detection methods.