PrognoseMES

Important data at a glance

Project title:	Development of an AI-based forecasting module for Manufacturing Execution System (MES) for predictive production control
Duration:	11/2019 – 01/2022
Collaborative project:	yes
Funding:	EFRE \| SAB
Processor / Contact person:	M.Sc. Mauritz Mälzer
Cooperation:	Seniorprofessur Prof. Dr.-Ing. habil. Klaus Kabitzsch, Institut für Angewandte Informatik (IAI), ccc software gmbh
Logline:	This project will develop a software add-on that integrates recent technological advances in machine learning into existing MES environments.

Objective

Introduction

The importance of MES systems as central repositories of production data in the manufacturing industry is growing steadily in the environment of increasing digitalization. A meaningful evaluation of the data enables robust production control, improved machine utilization, more efficient use of resources, and a more reliable status description of the entire production and operating resources in the production environment. Up to now, the evaluation of the accruing information has been carried out in the immediate machine environment by empirical values of long-standing employees or in the area of management by calculating performance indicators. The possibilities arising from recent advances in data analysis and in the field of "artificial intelligence" are still insufficiently embedded in conventional MES systems. This is where the PrognoseMES project comes in.

Objective

Supplementing existing analysis tools for processing production data with the possibility of using machine learning.
Creation of a platform for recording and evaluating industry-relevant news from the WWW.

Solution

Learning algorithms require cleanly prepared data sets, continuous adaptation to new data points, and trained evaluation of prediction results. The joint development should result in a software add-on, which can be provided as a configurable solution and adapted by the using company to its own needs. The data analysis results as a guided workflow of partially automated single steps. Of considerable importance is the interpretability for the end users. In the management area, the reliability of the prediction plays a major role; at the machine level, concrete cause-effect relationships are of interest. The traceability of a prediction to the main responsible input variables is decisive for the acceptance of the solution and should therefore be given special attention.