CYCLOSTATIONARY PROCESSES: APPLICATION IN GEAR FAULTS EARLY DIAGNOSIS

The paper introduces the theory of cyclostationary processes as a powerful tool for the diagnosis of rotating machines. It is shown that the well-known synchronised averaging method is the expression of cyclostationarity at the first order, and that the spectral correlation function issued from the second-order cyclostationarity is an efficient parameter for the early diagnosis of faults in rotating machines. Moreover, it is shown that vibration signals measured on gear systems display second-order cyclostationarity. Application in early diagnosis of spalling in gear teeth demonstrates the power of this new parameter.