Device, method, and computer program for screening cathode active material candidates for secondary batteries

Screening device for selecting cathode active material candidates by optimizing training and validation ratios

This technology optimizes the ratio of training data to validation data based on dataset performance evaluation results and uses predictive modeling for substitute particle combinations within specific crystal structures (layered structures) to automatically select high-performance cathode active material candidates.

Developing new cathode active materials is costly and time-consuming, and there have been technical limitations in identifying materials that minimize cobalt content while simultaneously achieving high energy density and structural stability.

This technology receives labeled datasets, determines training data ratios through model performance evaluation using validation datasets, and generates predictive models to determine substitute particle ratios within specific compositions (Chemical Formula 1) and layered structures. By screening candidates in this manner, it can effectively contribute to securing commercial competitiveness for lithium secondary battery cathode materials.

‍

‍

Key Features:

• Database component that receives datasets labeled with the properties of cathode active material structures for secondary batteries
• Preprocessing unit that processes a portion of the received datasets into training datasets for model learning
• Predictive model generation unit that predicts performance indicators for target materials to be placed in the established structure based on the training dataset
• Configuration that optimizes the ratio of training data to validation data by evaluating the results of the cathode active material predictive model

‍