Mining Shop-Floor Data for Preventive Maintenance Management: Integrating Probabilistic and Predictive Models☆

Abstract Well-defined intervals between preventive maintenance inspections help increase the availability of process equipment. This work integrates probabilistic and predictive models constructed from event logs information, through process mining techniques, to estimate the variations of process cycle time. A function for equipment availability is defined and gradual changes in intervals between maintenance inspections are performed until the wasted time for this function is minimized and the best interval is found. Different scenarios can be analyzed simulating variations in probabilities of activities occurrence, providing better support in scheduling maintenance activities. Results show that losses can be reduced by optimizing the intervals between maintenance inspections.