Hybrid scheme for wind turbine condition monitoring based on instantaneous angular speed and pattern recognition

Abstract

Wind turbines are designed to operate under varying conditions of speed and load. These rough operational conditions undermine conventional monitoring techniques and lead to unexpected failures of mechanical components. This work comes within the framework of wind turbine on-line condition monitoring. For this purpose, a particular attention was given to Instantaneous Angular Speed (IAS) emerging as a viable alternative to vibration analysis, especially in non stationary conditions. In this work, IAS signals were recorded from extensive measurement campaigns on different operating wind turbines. Suitable processing techniques have been specifically developed and allowed to analyze signals in healthy condition and in the presence of different bearing faults. Based on the latter, a huge number of expected relevant indicators was extracted. Different configurations of features transformation, selection and classification tools were tested. An optimized hybrid scheme has been designed. This approach allowed an optimal exploitation of IAS information and the construction of an effective tool for wind turbine condition monitoring.