Next generation of industry such as those promoted by “Industry of the Future”, “Industry 4.0.” paradigm, holds the promise of increased flexibility/adaptability in production (e.g. manufacturing) to cope with the challenges of producing individualised products as expected by customers with a short lead-time to market and at the cost of mass production. These challenges can only be met by further developing the digitalisation of production systems (integration of physical and IT-based digital worlds) as defended by new manufacturing concepts, such as “Biologicalisation in Manufacturing” or “Cloud Manufacturing” in which data science, smart manufacturing objects (SMO) and services are predominant. This vision should increase global competitiveness by promoting innovative business models mainly driven by servitization (anything-as-a-service) and stakeholders’ collaboration in the way to keep industrial employment in Europe. It led to the introduction of new technologies/techniques like IoT, advanced ICT, Big Data Analytics (BDA), cloud computing and Cyber-Physical Production System (CPPS). More precisely CPPS is a new way to organise the system (heterarchical vs holistic approach) to enable fast integration feedback and control loops throughout distributed manufacturing infrastructures all along its life cycle. So, the resulting organisation is assimilated to a complex manufacturing ecosystem based on interaction of humans, objects (e.g. products, components), customers, society partners … having to offer a dynamic, real-time optimized and self-organizing value chain.
These advanced characteristics imply that the system in support of the value chain need to become more adaptable, agile, robust, resilient … to face fault, unforeseen events while guaranteeing system performance. However, the process of predicting reliability and performance in such context is far from trivial. The main barriers include, at least, the inability to anticipateunknown faults particularly for complex systems, the inability to sustain system functionality and performancein the presence of system anomalies and severedisturbances and the inability to self-adjust system configurations to mitigate internal faults or external intrusions to achieve survivability.To face these barriers, the conventional challenges on diagnostics, prognostics, Fault-Tolerant Control, Maintenance … should be extending, mainly by using Artificial Intelligent tools to construct concepts such self-healing, self-assessment, self-maintenance, self-repair … as promoted by the Prognostics and Health Management (PHM) community.
Thus, the goal of this keynote speech is to provide insights into various aspects of these new concepts within the frame of “Industry of the Future”/CPPS and to discuss directions for future research on these concepts. .
Prof. Benoît IUNG, University of Lorraine-France
Benoît IUNG was born in 1962 in Nancy, France. He is full Professor of Prognostics and Health Management (PHM) at Lorraine University (France). He conducts research at the Nancy Research Centre for Automatic Control (CRAN, CNRS affiliation, UMR7039) where he is co-managing today a research group on Sustainable Industrial System Engineering (about 55 people). His research and teaching areas are related to dependability, advanced maintenance engineering, prognostics, heath management, e-maintenance and cyber-physical production system (CPPS). In relation to these topics he took scientific responsibility for the participation of CRAN in a lot of national, European (i.e. REMAFEX, DYNAMITE) and international projects, for example, with China (i.e. EIAM-IPE, CENNET) and Chile (i.e. iMaPla). He has numerous collaborations with industry in the frame of Convention for Research Program (mainly in France with EDF, CEA, RENAULT) and serve as responsible of a common Lab called PHM-FACTORY with PREDICT SME (ANR LabCOM). He is the chairman of the IFAC TC5.1. (2017-2020).
He was until 2014 the chairman of the IFAC WG A-MEST on advanced maintenance, and until 2018, the chairman of the ESRA TC on Manufacturing. He is a CIRP fellow since 2017, a PHM Society Fellow from 2018, a founding Fellow of ISEAM and of the European IAM Academic and Research Network, a nominated member of the IFAC TC 5.3. He had also a guest position in the NSF industry/university cooperative research centre for intelligent maintenance system (Univ. of Cincinnati; Pr. J. Lee, until end of 2017). He serves as correspondent on “Factory of the Future” for the University of Lorraine since 2016. Benoît Iung has (co)-authored over 200 scientific papers (65 journal papers) and several books including the first e-maintenance book in Springer. He developed several keynote speeches in international conferences and attended as reviewer, faculty opponent or examiner for a lot of Ph. D. and “docent” defenses in France and in Europe (UK, Belgium, Netherlands, Norway, Spain, Sweden, and Italy). He has supervised until now 18 Ph. D. Students and more 25 MA.
He served as IPC member of various IEEE and IFAC conferences and developed expertise/reviewing for European Commission in the frame of H2020 from 2015. He is a senior Editor of the new IFAC Journal of System and Control (Elsevier). Since 2014, he is the treasurer of the IEEE Reliability French chapter. Benoît IUNG received his B.S., M.S. and Ph.D. in Automatic Control, Manufacturing Engineering and Automation Engineering, respectively, from Lorraine University, and an accreditation to be research supervisor (2002) from this same University.
Security and cost reduction imperatives are increasingly driving industries towards conditional predictive maintenance.
It is made possible by acquisition of many data related to the integration of different sensors within the systems. The areas of statistical learning in general and deep learning in particular are now sufficiently mature to allow the development of technologies with artificial intelligence that can better anticipate technical failures than humans. This is true provided you have enough data to program it.
This talk will introduce the principles of deep learning and show how to use it for diagnosis (static and dynamic), to manage multimodality, detection of change, outliers, and weak signals analysis.
Prof. Stephane Canu, INSA Rouan-France
Stéphane Canu is a Professor of the LITIS research laboratory and of the information technology department, at the National institute of applied science in Rouen, Normandy.
He has been the dean of the computer engineering department he create in 1998 until 2002 when he was named director of the computing service and facilities unit.
In 2004 he join for one sabbatical year the machine learning group at ANU/NICTA (Canberra) with Alex Smola and Bob Williamson. In the last five years, he has published approximately thirty papers in refereed conference proceedings or journals in the areas of theory, algorithms and applications using kernel machines and deep learning.
His research interests includes deep learning, kernels machines, regularization, variable selection and optimization in machine learning.
The Power-to-X (P2X) technology referring to a number of electricity conversion and energy storage is an innovative approach and can be used to address the intermittency of renewable energy sources, such as wind and solar, and to balance network loads during average and peak electricity demand. Moreover, it has a real potential to contribute not only regarding reduction of greenhouse but also utilization of surplus energy across different industries.
From a practical point of view, the objective of the plenary speech is to present an industrial methodology for implementation of Power-to-X technologies by developing innovative indirect of electricity to fuels (hydrogen) from green sources and direct conversion of electricity with CO2 to platform chemicals developed in the framework of an European project.
From fundamental academic side, supervision (in terms of diagnosis and optimal operating modes management) of such process is complex because of its hybrid aspect due of interconnection between different varieties of sources and storage units, where each or some can be connected and disconnected, leading to a dual discrete-continuous dynamical behavior.
The present speech concerns those two complementary problematics to be discussed (i) state of art and problematics of P2X technologies based on real industrial (ii) literature synthesis of supervision of Hybrid Dynamic Systems(HDS) and use of new formalism named Even Driven Hybrid Bond Graph (EDHBG) for online supervision of Hybrid Renewable Energy Systems HRES applied to a real multisource platform submitted to intermittent sources.
Prof. Belkacem OULD BOUAMAMA, Polytechnique of Lille-France
Belkacem OULD BOUAMAMA is full Professor, and head of international relations and research at « Ecole Polytechnique Universitaire de Lille, France) » He is the leader of Bond Graph group at the CRIStAL (Research center in Computer Science, Signal and Automatic Control of Lille), Laboratory of the National Center for Scientific Research, where his activities concern Integrated Design for Supervision of System Engineering.
Their application domains are mainly ITS, nuclear, energy, and mechatronic systems. He is the author or coauthor of over 100 international publications in this domain and co-author of four books in Fault Detection and Isolation, mechatronics, bond graph modeling and Intelligent Transportation Systems.