Architecture for Predictive Maintenance Based on Integrated Models, Methods and Technologies

The evolution to Industry 4.0 is creating the impetus for the manufacturing industry to increase productivity through smart management and stabilization of resources, capacity and utilisation. Increased plant availability, extended service life of resources as well as optimised product and process quality require intelligent maintenance strategies. The conventional reactive maintenance (run-to-failure) causes unexpected production stoppages, and preventive maintenance at times leads to waste of working hours and material due to the premature replacement of machine components. A smart Predictive Maintenance (PdM) strategy equipped with fault detection and prediction based on acquired, processed and analysed data can result in an accurate estimation of the Remaining Useful Life (RUL) of machine components and thus trigger appropriate maintenance action plans. Data acquisition, processing, analysis and rule-based decision supporting require the development, application and combination of various Industrial Internet of Things (IIoT) devices, models and methods in an integrated manner. Through transparent development and integrated harmonisation of all models, methods and technologies, fault detections and respective RUL estimations of machine components become more accurate and reliable. This leads to an increasing acceptance of employees towards software-based recommendations, in particular maintenance instructions for operators and proposals for an optimised development of the next generation of production systems and equipment. Within the scope of the EU-funded project Z-BRE4K, this paper proposes an IIoT architecture that presents models, methods and technologies in an integrated manner and highlights the data and information flow between them. The architecture including the infrastructure has been applied in three pilot cases with the industrial end users PHILIPS, GESTAMP and CDS to demonstrate the compatibility of the architecture to different industries with various production systems and diverse conditions, requirements and needs. Based on the adaption of the generic architecture for the pilot cases, the models, methods and technologies were developed efficiently and continuously improved and validated. The proposed architecture is intended to be applicable across industries to facilitate the transformation from reactive or preventive to PdM and thereby improve the competitiveness of manufacturing companies.