Predictive Manufacturing Tardiness Inference in OEM Milk-Run Operations

Abstract

In an OEM milk-run pickup operation over a road network, the manufacturing of components by suppliers is subject to varying tardiness levels on order release dates. Such faults are traditionally diagnosed and treated with a “fail and fix” strategy (FAF), when a failure is recognized as a sudden disruption problem. In practice, quite often a degradation phase occurs in the manufacturing process before a disruption happens. But, within the Industry 4.0 paradigm, it is necessary to prevent faults that may occur at some time in the future, changing the traditional FAF response to a robust predicting and preventing strategy. In such a context, faults must be forecasted in a dynamic way, over a Big Data basis, and the resulting forecasts must be released at once to the logistics agent to allow him to review his milk-run collecting program in due time, thus leading to a better integrated performance. An approximate method to forecast tardiness levels in supplier’s production, intended to help the related logistic operators to reschedule their services in due time, is proposed and illustrated with a case study.