Quality in production planning: Definition, quantification and a machine learning based improvement method

Abstract

The reliability of production plans drops drastically within several days after plan creation. The reasons for the deviation between planning and execution are manifold. Causes can be e.g., uncertainties, inaccurate or insufficient planning data (e.g., data quality and availability), inappropriate planning and control models and systems or unforeseeable events, leading to high control effort in order to reach the desired logistical KPI's, such as due date reliability or lead time. The paper addresses this problem with machine learning and states a methodological approach to measure and improve the quality in production planning. Although the term “planning quality” (PQ) has been used several times by the scientific community, a standard definition of PQ in the field of production planning and a clear distinction to other similar concepts like “robust planning” is still missing. Furthermore, PQ and its application within the production planning process are evaluated with a real industrial use case from the steel manufacturing industry.