Periodic Reporting for period 2 - SYNERGY (Synergy for Smart Multi-Objective Optimisation)
Okres sprawozdawczy: 2017-05-01 do 2019-01-31
Many real-world application areas, such as advanced manufacturing, involve optimisation of several, often time-consuming and conflicting objectives. They may require the maximisation of the product quality while minimising the production cost, relying on demanding numerical simulations. These multi-objective optimisation problems can be solved more effectively if parallelisation is used to execute the simulations simultaneously and if the simulations are partly replaced by surrogate models. The overall goal of the SYNERGY project is to enhance the initial-stage research in multi-objective optimisation at the Jožef Stefan Institute (JSI), Slovenia. Here, JSI collaborates with two high-profile research institutions with complementary expertise, Université des sciences et technologies de Lille (USTL), France, and TH Köln - University of Applied Sciences (CUAS), Germany. While USTL leads parallelisation on large-scale heterogeneous architectures, CUAS provides expertise in surrogate modelling in multi-objective optimisation.
The project overall goal of reducing JSI’s research networking gaps have been achieved through three core objectives aimed at increasing the number of joint papers with the high-profile institutions, increasing the number of citations, continued/expanded collaboration with the high-profile research institutions after the end of project and enlarged JSI’s visibility for industry, in particular in Slovenia.
The SYNERGY project significantly strengthened JSI’s expertise in surrogate-assisted multi-objective optimisation and large-scale parallel architectures for multi-objective optimisation, and provided the support actions needed to merge these two efficient concepts into smart multi-objective optimisation. Together, the partners were able to deal with the optimisation of complex real-world problems more efficiently and disseminated the acquired knowledge to Slovenian industry and other research institutions. The results of this collaboration raised the research profile of JSI staff, supported excellence of USTL and CUAS and provided means for spreading innovation to the industry. In addition, SYNERGY’s underlying topic of smart multi-objective optimisation contributed to the Smart Specialisation Strategy of the Republic of Slovenia, which defines Smart factories as one of its priority areas.
The project overall goal of reducing JSI’s research networking gaps have been achieved through three core objectives aimed at increasing the number of joint papers with the high-profile institutions, increasing the number of citations, continued/expanded collaboration with the high-profile research institutions after the end of project and enlarged JSI’s visibility for industry, in particular in Slovenia.
The SYNERGY project significantly strengthened JSI’s expertise in surrogate-assisted multi-objective optimisation and large-scale parallel architectures for multi-objective optimisation, and provided the support actions needed to merge these two efficient concepts into smart multi-objective optimisation. Together, the partners were able to deal with the optimisation of complex real-world problems more efficiently and disseminated the acquired knowledge to Slovenian industry and other research institutions. The results of this collaboration raised the research profile of JSI staff, supported excellence of USTL and CUAS and provided means for spreading innovation to the industry. In addition, SYNERGY’s underlying topic of smart multi-objective optimisation contributed to the Smart Specialisation Strategy of the Republic of Slovenia, which defines Smart factories as one of its priority areas.
Training activities focused on high-performance computing and optimisation as well as surrogate-assisted multi-objective optimisation were dedicated to JSI researchers, Slovenian students and other researchers interested in the topics of the SYNERGY project. High-performance computing training included lectures and a practical lab exercise. They were dedicated to the didactical presentation of a unified view of metaheuristics for multi-objective optimisation, getting started with practical training on the Grid’5000 experimental testbed, and implementation of parallel algorithms for solving challenging problems on large GPU-powered clusters. Training activities with focus on surrogate-assisted multi-objective optimisation consisted of introduction to different modelling technique, with special attention to different interactions with surrogate-models and its applicability to real-world optimisation problems. A wider audience was addressed through several invited lectures.
Consortium members have performed many activities related to a variety of events aimed at sharing the existing knowledge and gaining new knowledge on smart multi-objective optimisation. These included from chairing international workshops, organising special sessions and conference tracks to organising several workshops, conferences, a summer school and also an industrial challenge and job market sessions.
The most prominent organised events were the International Conference on Bioinspired Optimisation Methods and their Applications, BIOMA 2018, that took place in Paris, France, during 16–18 May 2018 (http://bioma2018.sciencesconf.org) and SYNERGY Summer School on Efficient Multi-Objective Optimisation that took place in Ljubljana, Slovenia, during 27–31 August 2018 (http://summerschool.synergy-twinning.eu). BIOMA conference proceedings were published by Springer in the renowned series Lecture Notes in Computer Science (LNCS).
Dissemination and outreach were performed through two tutorials, five invited lectures, a journal special issue, three invited seminars, two keynote talks, chairing a track and serving in two steering committees. Audiences included participants of international scientific conferences as well as Slovenian events, which aimed to directly address national stakeholders.
Results of the improved excellence of JSI researchers are evident from several tutorials given by JSI researchers at GECCO and CEC conferences, and successful proposals for a workshop and a special session at CEC 2019 without the influence of researchers from the high-profile partners.
Smart specialisation stakeholders have been contacted on many occasions and the most prominent achievement was co-organisation of the event with the Chamber of Commerce and Industry of Slovenia (GZS) where successful applications of research achievements to real-world problems were presented to their interested members.
Promotion of women in science was performed by presenting research work to the general public during the JSI Open Day, co-organisation of the Women@GECCO workshop and a round table Women in Science and Technology at the Jožef Stefan Institute.
Consortium members have performed many activities related to a variety of events aimed at sharing the existing knowledge and gaining new knowledge on smart multi-objective optimisation. These included from chairing international workshops, organising special sessions and conference tracks to organising several workshops, conferences, a summer school and also an industrial challenge and job market sessions.
The most prominent organised events were the International Conference on Bioinspired Optimisation Methods and their Applications, BIOMA 2018, that took place in Paris, France, during 16–18 May 2018 (http://bioma2018.sciencesconf.org) and SYNERGY Summer School on Efficient Multi-Objective Optimisation that took place in Ljubljana, Slovenia, during 27–31 August 2018 (http://summerschool.synergy-twinning.eu). BIOMA conference proceedings were published by Springer in the renowned series Lecture Notes in Computer Science (LNCS).
Dissemination and outreach were performed through two tutorials, five invited lectures, a journal special issue, three invited seminars, two keynote talks, chairing a track and serving in two steering committees. Audiences included participants of international scientific conferences as well as Slovenian events, which aimed to directly address national stakeholders.
Results of the improved excellence of JSI researchers are evident from several tutorials given by JSI researchers at GECCO and CEC conferences, and successful proposals for a workshop and a special session at CEC 2019 without the influence of researchers from the high-profile partners.
Smart specialisation stakeholders have been contacted on many occasions and the most prominent achievement was co-organisation of the event with the Chamber of Commerce and Industry of Slovenia (GZS) where successful applications of research achievements to real-world problems were presented to their interested members.
Promotion of women in science was performed by presenting research work to the general public during the JSI Open Day, co-organisation of the Women@GECCO workshop and a round table Women in Science and Technology at the Jožef Stefan Institute.
With the gained knowledge, applied to real-world problems in Slovenian industry, it is expected that innovative solutions will be developed to efficiently solve the presented problems. In particular, with Štore Steel d.o.o. a project to support their steel production with a simulation-based multi-objective optimiser, and with INEA d.o.o. where scheduling of flexible offers for energy production and consumption is carried out with the goal of minimizing the mismatch between the two. With Domel d.o.o. we are working on designing their latest product with the application of advanced multi-objective optimisation approaches. Great possibilities for our research to be used in Smart Specialisation were also noticed by the Chamber of Commerce and Industry of Slovenia, with several discussions with companies are ongoing. We believe collaboration shows a great potential to propel Slovenian industry closer to the smart factory of the future. Through giving lectures at the faculties and institutes we believe we have reached a wider audience from students to other researchers who were acquainted with applying parallelisation and surrogate modelling to multi-objective optimisation.
Economic impact is not achieved only by the JSI. USTL has initiated industrial collaborations using the synergy between multi-objective optimisation, surrogate modelling and high-performance computing to solve complex multi-disciplinary design engineering problems in aerospace (e.g. Launcher design). This work is performed in collaboration with ONERA (French Aerospace Lab) and CNES (National Institute of Aerospace Studies). Other complex engineering problem in smart grids (collaboration with EDF) have also been initiated. Similarly, CUAS established or intensified several contacts to German partners. In addition, since January 2017 JSI and CUAS collaborate in another EU project, Innovative Training Network (ITN) named UTOPIAE.
Economic impact is not achieved only by the JSI. USTL has initiated industrial collaborations using the synergy between multi-objective optimisation, surrogate modelling and high-performance computing to solve complex multi-disciplinary design engineering problems in aerospace (e.g. Launcher design). This work is performed in collaboration with ONERA (French Aerospace Lab) and CNES (National Institute of Aerospace Studies). Other complex engineering problem in smart grids (collaboration with EDF) have also been initiated. Similarly, CUAS established or intensified several contacts to German partners. In addition, since January 2017 JSI and CUAS collaborate in another EU project, Innovative Training Network (ITN) named UTOPIAE.