Databased prediction and planning of order-specific transition times

Abstract

An increasing demand for individualized products leads to smaller lot sizes and a shift towards workshop manufacturing. Similarly, customers demand for cost-efficient production, high delivery reliability and short lead times, which are difficult to achieve in a workshop production at once. As a result, transition times, the times between value-adding operations, can account for more than 90 % of lead times, but are often treated as static master data and still estimated manually. In this article, resulting prediction improvements for order-specific transition times based on methods of machine learning are presented and an approach for integration into practice is discussed.