Application of cointegration to vibration signal for local damage detection in gearboxes

Abstract

The problem of local damage detection for rotating machines is widely studied in the literature. In order to extract the information of the damage, the approach based on the vibration signal analysis is usually applied. Very often classical methods suited for impulsiveness analysis are not sufficient, therefore it is proposed to analyse the signal in terms of its periodicity features. In this paper the cointegration approach is applied to vibration signal in the context of local damage detection in gearbox. The goal of this approach is to recognize if the examined signal comes from healthy or damaged machine. Firstly, we assume periodic correlation of given signal and measure its period. Afterwards, signal is restructured and divided into sub-signals according to the discovered period. Finally, we check if sub-signals are integrated and calculate the cointegrating vector by using the least squares method. In the paper we present details of the method and benefits of using our procedure. To validate our methodology we used simulated vibration signal based on typical gearbox used in mining industry, as well as real data from the gearbox. In the paper authors present a concept of the complete procedure that uses cointegration interpretation as a diagnostic measure.