The Level-Up project has wrapped up with its Final Conference
Authors: Dimitris Eleftheriou, Ioannis Meintanis, Yianna Sigalou
17th October 2023
After four fruitful years, the Level-Up Project has wrapped up with its final deliverables and a Final Conference held in Brussels, Belgium. The project aimed to develop a platform that extends the useful life of major capital investments.
As part of the project, CORE Innovation Centre developed a range of models using ML-based algorithms across different use cases. Our team was also responsible for exploitation and communication activities for the Consortium.
The aim of Level-Up was to offer a scalable platform covering the overall life-cycle of critical role, big industrial machines or their components, starting from the initial digital twins setup to facilitate predictive maintenance, modernisation actions to diagnose and predict the operation of physical assets, even to the refurbishment and re-manufacturing activities towards the end of a machine’s life.
Different machine learning models
The machine learning (ML) based tools developed by our team have been applied to 4 different manufacturing lines (ESMA, LUCCHINI, TOSHULIN, and IPC) at the component, work-station & shopfloor level, with different technologies used depending on the specific needs, and available data for each pilot line.
For ESMA’s cold forming press, our team implemented AI based anomaly detection (AD) algorithms using state-of-the-art Deep Learning (DL) architectures, such as Auto-Encoders and proprietary unsupervised learning algorithms. By utilizing vibration signals and a variety of IoT sensors placed in the equipment, the models can look for patterns in data that indicate failure modes for specific components (e.g. bearings) and provide insights in real-time for the machine operator.
The TOSHULIN production line consists of a large industrial vertical lathe (SKIQ16-v2), with the workpiece clamped on a clamping plate which rotates when in operation. The end-user requirement was to focus on the lubrication system of the cutting tool, to detect anomalies and assess its operation capabilities. To achieve this, we developed a combination of tools, which utilizes a forecasting model to predict the future machine states and the behavior of the oil particles, together with a flexible monitoring mechanism which utilises dynamic thresholds to detect anomalies.
LUCHINI is a full production line for machining railway axles, and for Level-Up we developed an AD procedure using multi-sensorial vibration data. The goal was to facilitate predictive maintenance for the two most critical machines of the production line. The AD procedure is currently at the on-line/production stage, and we continue to monitor the performance and accuracy of the models used.
For IPC/CRF’s pultrusion pilot line, machine learning algorithms for AD and quality control have been developed and integrated with their upgraded monitoring dashboard to assist the operator in decision making and process monitoring procedures.
A go-to-market pathway for the consortium
Our team also led on exploitation activities for the project, to maximise the impact of its results. We developed a detailed exploitation plan for 26 of the project results, across 6 different sectors for our Consortium partners. For each use case, a detailed business plan was developed, which included:
Innovation Management Activities: We analysed the external ecosystem through which Level-Up can evolve, using different strategic tools, like SWOT and Porter’s Five Forces. We analysed the market for each sector, as well as potential market barriers that might slow the adoption of the technologies developed.
Business Models: For each use case, we developed a detailed business model using the Business Model Canva tool, identifying unique selling points, customer personas, costs, potential revenue streams, and key go-to-market activities.
Exploitation Roadmap: We developed detailed exploitation roadmaps and commercialization analyses for the project technologies, accompanied by a 5-year financial plan, which partners can use as a reference in their go-to-market journey.
Reaching out to the community
Finally, CORE Innovation Centre was responsible for handling the communication and dissemination activities of Level-Up.
The project excelled on social media, attracting 904 followers on Twitter and 981 on LinkedIn. During the project, we created 22 videos overall, which collectively received over 3,600 views.
With support from our partners at AIMEN and Innovalia, we carried a series of summer workshops to showcase the project's final results. We also co-organized the project's final conference in Brussels, with more than 70 people in attendance.
Through its four-year run, our Consortium submitted over 7 papers in open access journals, with 4 more pending approval for publication, and attended over 60 industry or scientific events.
