Computer aided diagnostic methods to forecast condition-based maintenance tasks

Abstract

Nowadays, production and service processes rely heavily on computerized supervision. The data provided by these supervisory systems are also crucial for decision-making, involving maintenance and revision. However, these data can include various measurement and sampling errors. Taking measurement uncertainty into consideration the loss can be decreased. In our model the realized (or lost) revenue is treated along with the occurred costs, in order to maximize the profit with the help of analytic calculation and simulations. This paper presents that by treating measurement results as a time series the time of probable failure can be predicted at a given confidence level. The uncertainty of measurement and uncertainty of forecasting can be treated in the same model. © 2013 Springer Science+Business Media.