Abstract

Maintenance in industry is currently moving from time planned preventive methods to condition-based operation for better process reliability and lowered manufacturing costs. Machine vibrations include information from operating state and machine health and can be used in the computing of several different features for condition monitoring and process control. These describing values can be used for the estimation of remaining useful life (RUL). Local computing enables the use of advanced algorithms for dense vibration data on-site, right next to the monitored process so that the data can be turned into information without the need for large data transfers and centralized computing. Calculated features can be supported with other sensory data, information through expert knowledge, modelling, and data from similar systems in other installations. Developments in wireless technologies enable the use of small nodes in distributed computing. This paper examines the use of locally calculated generalized norms in combination with supporting information from the global maintenance database.