14 Dec Artificial Intelligence makes manufacturing smart
Big data and artificial intelligence have the potential to provide the EU with the efficiency it needs to be competitive. With their development, comes new skills, new jobs and opportunities for economic growth.
Smart manufacturing projects, like PROTEUS, are leading the way in this area by incorporating the real world demands of the industry while providing an opportunity to test machine-learning models in an operational scenario.
On the 23rd November 2017, Trilateral’s PROTEUS project and other initiatives jointly organised a workshop on Industrial Data Platforms for the Manufacturing Domain. At the workshop, attendees heard how EU funded projects are identifying challenges and developing big data solutions in smart manufacturing, in both industrial use-cases and research and innovation pilots.
The workshop was part of the European Big Data Value Forum (EBDVF), held in Versailles, France, with a focus on “Trusted AI in Smart Industry“. The conference provided an opportunity for academics, industry representatives and policy makers to debate the obstacles and pathways for data-led innovation and investment in Europe.
During the workshop, we heard presentations in various areas of application:
- The PROTEUS coordinator, Marcos Sacristán (Treelogic, Spain), described an industrial use-case in steel manufacturing, a strategic industry in the EU economy.
Here, data collected by sensors is fed into machine learning algorithms to predict defects in the production of steel coils, thus allowing faulty processes to be modified at an early stage preventing the waste of significant resources. PROTEUS is also developing interactive visualization tools to inform manufacturers about the state of crucial quality control parameters including temperature, vibration, roller tension and speed.
- As one of the PROTEUS partners, we at Trilateral are developing the key performance indicators and benchmarks including the benefits to the steel industry and the EU’s big data capabilities more broadly. PROTEUS will build on the machine learning functionality of Apache FLINK, a distributed high-performance streaming framework popularized in Europe, by introducing algorithms that can simultaneously operate on both data-at-rest (batch data) and data-in-motion (streaming data). Trilateral will look to establish benchmarks in the accuracy of the machine learning components and the data handling performance of PROTEUS’ core Apache FLINK components.
- Thomas Hahn (SIEMENS, Germany) reported on the challenging need for interoperability and standardization in digitalization. Establishing standards is crucial to strengthening competitiveness and establishing the creation of value from big data.
- Dr Aizea Lojo (IK4-Ikerlan, Spain) and Dr Iñaki Garitano (Mondragon Unibertsitatea, Spain) spoke on the issue of cybersecurity in three European manufacturing projects, MATIS, ARROWHEAD and CREMA. As the use of smart devices increases, people become more connected and cybersecurity becomes ever more salient. If the market value of the Internet of Things is to exceed one trillion euros by 2020 as the European Commission forecasts, it is imperative that people feel safe and secure.
- Prof Ernesto Damiani (CINI, Italy) and Dr Mariangela Lazoi (Università del Salento, Italy) described how the TOREADOR project is building a big data platform for predictive maintenance in Aerospace manufacturing. Predictive maintenance uses machine learning to understand the root causes of equipment faults and enables the industry to monitor settings and avoid predictable breakdown.
The conference recorded an average of 800 participants a day with over 1200 participants over three days.