Abstract

Relevant classification of the stationary operating conditions of wind turbines (WTs) aids in the selection of an optimal condition monitoring technique. This paper presents a general method that can be used to classify the operating conditions of WTs in terms of rotor speed and power. In this study, the ideal probability density functions (PDFs) of rotor speed and power are calculated using an analytic WT model and a wind speed profile. To estimate the PDFs of rotor speed and power with field data, two methods are employed: (1) empirical PDF-based and (2) Gaussian mixture model (GMM)-based. The individual PDFs estimated by the two methods are used to quantitatively define the range of the stationary WT operating conditions. The proposed methods and the range of stationary operating conditions established by the methods were evaluated using data from an analytical WT model and an actual 2.5 megawatt WT in the field. In addition, the paper presents the evaluation of the performance of the proposed class-wise condition monitoring strategy when used with vibration signals acquired from a two kilowatt WT testbed. In summary, the proposed strategy and methods are promising for effective condition monitoring of WTs.