Speed estimation in geared wind turbines using the maximum correlation coefficient

Abstract

Valid speed signal is essential for proper condition monitoring of modern variable speed wind turbines. Traditionally, a tachometer mounted on the high speed shaft provides reference for tracking speed dependant frequency components, such as generator speed harmonics and gearbox tooth mesh frequencies. The health assessment of drive train components is limited to broadband measurements when the speed signal is invalid. This condition results in reduced fault detection capabilities and consequently decreased lead time. In this work, a new speed estimation algorithm is presented in order to overcome the above mentioned issues. The high speed stage shaft angular velocity is calculated based on the maximum correlation coefficient between the 1st gear mesh frequency of the last gearbox stage and a pure sinus tone of known frequency and phase. The proposed algorithm utilizes vibration signals from two accelerometers for cross-referencing purposes. The method is tested in three drive train configurations, where 720 sets of vibration signals of 10.24s length, sampled at 25.6kHz are analysed. Consistent speed estimation reaches approximately 98% when two vibration sources are utilized, whereas it is lower when only one source is taken into account. No apparent patterns arise between speed variation levels or power production and the number of invalid outputs, showing the independence of the method from operational parameters.