Method of Analysing Non-stationary Electrical Signals

Considering the non-stationary operating conditions of wind turbines, electrical signals measured at the stator of their generators will also present variations around their fundamental frequency. This paper presents a method able to efficiently analyse electric quantities measured at the generator stator. The obtained outputs consist in electrical features that fully describe the electrical information contained in the measured three-phase quantities. These features can be directly used or further analysed to obtain efficient fault indicators. The proposed method relies on using the instantaneous symmetrical components to describe the quantities and complex-valued filtering to select the content around the fundamental frequency. The obtained sample per sample algorithm can be implemented on-line, and is able to process stationary or non-stationary quantities in order to extract the useful information around the fundamental frequency. The performance of the proposed method, as well as its capability to detect mechanical faults, is illustrated using experimental data.