As part of the GIGAGREEN project’s digitalisation roadmap, we’re proud to present the Cathode Quality Simulator, an interactive web-based tool developed to support lab technicians in optimizing cathode production parameters.
While not yet a full digital twin, this simulator represents a key first milestone toward that vision. Built using machine learning techniques, it offers predictive insights based on historical lab data, helping users understand how different process conditions affect cathode quality.
At its core is an XGBoost algorithm, a powerful model trained on carefully curated and balanced data. We addressed challenges like class imbalance (where good cathodes greatly outnumber bad ones) and streamlined feature selection to focus on the most relevant inputs, such as mixing speeds, slurry composition, coating settings, and drying temperatures. Less impactful variables are automatically fixed to median values, making the interface both accessible and efficient for technicians.
The results are promising: the simulator achieved a 95% prediction accuracy on new, unseen samples and includes confidence scores to guide decision-making. It’s already proving useful in revealing hidden patterns in process data and reducing trial-and-error in lab-scale manufacturing.
This tool sets the stage for future integration with real-time data and process control, a major step toward a true digital twin for cathode production. With this foundation in place, GIGAGREEN continues to lead the way in combining sustainability, innovation, and smart manufacturing.
