Rolling bearing condition monitoring based on frequency response analysis

This paper deals with a new diagnosis method for detecting bearing faults. The diagnosis is carried out based on frequency response analysis. The detection of bearing damages on motor and load side of a mechanical drive system is addressed. The mechanics are assumed to be a two-inertia-system with one dominant resonant frequency. The frequency response of the mechanical system of the drive is obtained by using two measured signals. This method is regarded to be superior to one channel analysis that can be found in many technical papers in this field of research. The measurement of the required signal can be accomplished during operation of the plant in closed loop speed control. The system is excited by pseudo random binary testsignals. The deviations between the frequency response obtained during the commissioning of the plant and the curve measured under fault condition serve as indicators for the damage. The present paper addresses the detection of outer and inner race bearing faults. The majority of technical papers often are dedicated to experimental research regarding outer race faults. As inner race faults are much more difficult to diagnose, they are quite seldom in the scope of investigations. Both types of faults cause characteristic fault frequencies, which influence the frequency response of the mechanics. This paper contains experimental results which point out the reliability of the proposed analysis concept for rolling bearing condition monitoring.