Instantaneous angular speed indicators construction for wind turbine condition monitoring

This paper deals with wind turbine condition monitoring in non-stationary conditions. It seeks the construction of relevant indicator databases for the optimal exploitation of Instantaneous Angular Speed information (IAS). The procedure is based on IAS signal processing tools, indicator transformation and Artificial Intelligence (AI) classification tools. Experimental signals were collected over long operating periods from healthy and defective machines. Suitable IAS processing techniques have then been specifically developed for the extraction of a first set of indicators. Based on the latter, two transformation approaches were applied for the generation of new indicator databases. To evaluate the performance of the new sets of indicators the Radial Basis Neural Network classification method was applied over the raw and transformed databases. These analysis allowed us to assess the effectiveness of the current approach.