Separation of Frequency Coexisting Vibration Signals in Wind Turbines Gearboxes

JAVIER ERNESTO KOLODZIEJ; GONZALO I. V. OKULCZYK; SERGIO MOYA; FERNANDO BOTTERÓN

Santa Maria - RS

Seminario; 11th Seminar on Power Electronics and Control SEPOC; 2018

Technology Center of the Federal University of Santa Maria

In this paper, a method for separation of vibration signals coexisting in the same frequency bin and coming from different mechanicals pieces is presented. A blind source separation method for vibration signals in gearboxes based on a clustering technique is applied. The proposed method uses the fact that spatial separation of vibration sources produces a difference in the hermitian angle between vectors composed by samples of the discrete Fourier transform of the signals measured and a reference vector. For simulations and a real case considering a wind turbine gearbox, time segments with signals providing from different mechanical pieces are identified.