Predictive and Reactive Scheduling for a Critical Machine of a Production System

Abstract

In the paper a production model with failures is presented where successive failure-free times are supposed to have normal distributions and are followed by normally distributed times of repairs. Unknown parameters of distributions are estimated using e.g. empirical moments approach. Predictions of unknown parameters are done using classical regression method. Having Mean Time To First Failure, and Mean Time of Repair a disturbance robust predictive schedule is generated using rule: Minimal Impact of Disturbed Operation on the Schedule. If a machine's failure-free time is not successfully predicted reactive schedule is generated using criterion: Minimal Impact of Rescheduled Operation on the Schedule.