Research Activities

Holding a big market share in the Balkan peninsula in the domain of offset printing, Pressious is exhibiting a significant environmental footprint, since the manufacturing process involves extensive usage of raw materials (water, paper, ink, aluminium), where defected products contribute the largest part. Although the environmental impact of the company and the problems associated with the product defects have already been identified and partially addressed, their current management leads to excessive cost. For this purpose, Pressious R&D team has already collaborated with various universities (NKUA) and research institutions in order to implement Machine Learning (ML) methods, taking advantage of the historical data that have been collected by the company across the production line (3-year historical dataset). Based on this sufficient and labelled dataset as well as the powers of ZDMP platform, the targets of the present activity primarily involve the minimization of the production chain defects and the reduction of the company’s environmental footprint, thus leading to the concept of zero-defect manufacturing (ZDM) and effective cost reduction/production enhancement. In parallel, this activity enables Pressious to provide an ecologically friendly and high-tech environment to test and validate 14 ZDMP components.

Under this prism, this project offers a unique opportunity for Pressious to establish Industry 4.0 principles including proactive resource management, environmental footprint minimization and production chain digitization by exploiting available datasets across the company’s manufacturing procedure. Additionally, Pressious aims at moving towards a new ecosystem, boosting ML solutions to the aforementioned problems. This activity aligns with ZDMP objectives by enabling Pressious to test and validate the efficiency of open-source tools. In particular, by using an extensive set of Pressious available data, the comprehensive validation of ZDMP platform and respective components will add significant value to the ZDM and, in general, to the Industry 4.0 era. It is worth noting that the real production chain of Pressious inherently involves all the validation/testing requirements of the open call, including not only functional, technical and scenario validation, but also performance and usability testing on specific ZDMP components. Last, but not least, embracing ZDMP, Pressious plans to significantly enhance already established circular economy aspects, thus highlighting the capabilities of ZDMP platform in a new industrial domain, namely offset printing.

In the framework of this project, real production scenarios stemming from Pressious production chain and addressing existing challenges will be fully evaluated. In this context, the validation of ZDMP platform and components will be carried out encompassing the aforementioned scenarios, thus combining functional, scenario and technical validation, as asked by the ZDMP Open Call. The uniqueness of our approach mainly relies on the cross-layer adaptation and interconnection of the ZDMP components within Pressious production chain. In particular, our methodology is based on labelled datasets obtained across several Pressious departments, incorporating both historical (last 3 years) and recently collected data and will allow accurate verification of the tools and comparison of the results with the actual historical events. Towards this direction, several ZDMP components will be fully evaluated (e.g. AI model prediction, security, alerting) in real production scenarios, as well as the InCloud usability of the ZDMP will be verified.

Another valuable outcome could be the extension of ZDMP existing AI algorithms used for anomaly detection, based on the interest of Pressious R&D personnel and expertise of NKUA pre- and post-PhD students.