Advancing Catalyst Recycling Empowered by Data-Driven AI/ML Predictive Tool
Under the framework of the FIREFLY project, INLECOM INNOVATION (Greece) develops a data driven Artificial Intelligence (AI) / Machine Learning (ML) tool to predict the operation and performance of electrochemical processes. This digital tool will help the users decide the most suitable chemical treatment to recycle individual samples of spent, off-specification or waste metal-based catalysts, by suggesting the optimal sequence of chemical processes considering constraints provided by the users.
During the initial six months of the project, INLECOM focused on identifying industrial challenges, reviewing relevant data for ML model training, and collecting user requirements – information much needed to develop a comprehensive predictive tool.
A major part of the design efforts for the AI/ML-based predictive tool lies in the elaboration and exploitation of appropriate amounts and types of data for algorithm training. INLECOM, in collaboration with the FIREFLY consortium dedicated significant efforts to clarify, structure and quantify the data required for the tool, with a strong emphasis on the end-users’ requirements.
Having completed the full system specifications of the predictive tool, during the forthcoming months INLECOM will dedicate efforts to two primary areas of focus. The first involves conducting in-depth studies to better understand the data and their preprocessing. The second one entails conducting experiments using selected ML algorithms on the preprocessed data to run an initial assessment of their performance.
Ultimately, during the final stages of the project, the tool will undergo validation in industrial use cases, which will confirm its efficacy and pave the way towards more sustainable and efficient recycling practices.