How can production data be utilised to predict and prevent machine failure? In the Leading-Edge Cluster it’s OWL, BENTELER and Fraunhofer IEM are working together on a pilot factory where predictive maintenance methods and tools are implemented in order to answer this question. The research collaboration in OstWestfalenLippe is part of the EU project BOOST 4.0 – Big Data for Factories.
The aim is to establish and roll out a predictive maintenance framework at the BENTELER factory in Paderborn. Moreover, a standardized process from data acquisition to integration within the maintenance process is implemented. By systematically collecting and analysing machine data of a hydraulic press system and a scrap belt, it will be possible to derive patterns in the production process at BENTELER. Fraunhofer IEM is developing a process model for the implementation of predictive maintenance, i.e. the upkeep and inspection of machines in order to prevent machine failure. Here, data-driven modelling methods (machine learning methods) will be utilised and developed further, thus making it possible to detect machine errors long before they occur. Breakdowns can thus be avoided, so as to decrease downtimes, for instance, or to improve the efficiency of the manufacturing process.
Currently, BOOST 4.0 is the largest European initiative for Big Data in industry. 50 partners from 16 countries are working on a European standard for Big Data which pay into the construction of the European Industrial Data Space. German reference architectures such as RAMI 4.0 or initiatives such as International Data Spaces will be combined with different methods and models to bring numerous communities together in all of Europe. Project results in industrial data analysis and data usage will be tested and utilised at 10 European locations, e. g. Spain, Italy, Sweden and Germany.
A first paper has been published out of the project: "Detecting Anomalous Behavior Towards Predictive Maintenance" was written by Dr. Daniel Köchling, BENTELER Automotive, Dr. Athanasios Naskos and Dr. Ifigeneia Metaxa, both ATLANTIS Engineering, as well as Dr. Anastasios Gounaris, Aristotle University of Thessaloniki.
A key Industrie 4.0 element is predictive maintenance, which leverages machine learning, IoT and big data applications to ensure that the required equipment is fully functional at all times. In this work, we present a case study of smart maintenance in a real-world setting. The rationale is to depart from model-based and simple rule-based techniques and adopt an approach, which detects anomalous events in an unsupervised manner. Further, we explore how incorporation of domain knowledge can assist the unsupervised anomaly detection process and we discuss practical issues.
The complete article is available at Springer International Publishing.