Automated Data Gathering, Classification, Standardization and Evaluation Connected with AI in Smart Li-ion Battery Cell Production

Abstract

The production of lithium-ion batteries is characterized by high scrap rates. With material costs of 70-80% of the cell price, these are a significant cost factor. By using AI, correlations of process data and cell properties can be determined and particularly quality-relevant process steps can be identified.However, the prerequisite for the use of AI is a clean collection, description and allocation of data which is very time-consuming.In the BMBF-funded projects “KontElPro”, for solvent-free electrode production and “AgiloBat2”, for the agile production of format-flexible cells, software is being developed to solve this problem.The corresponding software system "Batalyse" is capable of automatically collecting, categorizing, linking and evaluating test, analysis and production data. Quality-relevant parameters of the produced (lab) cells such as capacity, resistance and power can thus be directly correlated with process parameters and data. The complete process chain of input materials, (intermediate) products, sensors, equipment, process parameters and all evaluated result data of the produced cells is stored in the database. All data, correlations and logic is provided by this database via an interface for AI applications.The aim is to use this data to generate digital twins of each individual process step and to optimize them step by step. Later, the processes are to be controlled autonomously by AI applying optimized process parameters to avoid or at least to reduce scrap.In the talk we present the structure, status and use of the software for research data management and in the use of in the research production of Li-ion cells.