Signal analysis for machine failure detection and prediction

Abstract

This paper describes the development and implementation of a system to predict and detect failures in rubber covered steel rolls used in paper mills. This system is based on the analysis of machine vibrations using DSP techniques. The vibration signal is detected by accelerometers placed on the pressure rolls. It was found that simple signal analysis can easily detect and predict the onset of a machine failure.