Improving KPI Based Performance Analysis in Discrete, Multi-variant Production

Abstract

Discrete production systems face the challenge of moving from a mass to a mass-customised production model. Classic methods for analysing Key Performance Indicators (KPI) that are based on a statistical approach are difficult to apply in the case of short series, multi-variant production. A new approach for KPI analysis that is based on machine learning and data mining methods has to be applied. The authors propose a new approach that is based on K-means clustering that can be useful for performance analysis in the case of short series, multi-variant production. The presented research is focused on discrete production systems with KPI data traceability on the work cell level. The main advantage of the presented solution is its ability to automatically estimate a number of technological variants that affect a given performance indicator.