Predict failure time in a mechanical system while the system is running

The application of condition monitoring techniques within industry to manage maintenance actions has increased swiftly over recent years. Knowing when a special component fails is very important because by replacing it before the failure time, many troubles such as loss production, injuries because of failures, damaging other components, endangering the safety of factories can be avoided. In this paper, a decision making model based on condition monitoring big data is proposed. This model uses condition monitoring big data as covariates and finds their effects on the life time of a component. Proposed model is first trained offline using previously recorded logfile data, and is then used to predict failures online – while the system is running. .