• Goal

- Sequence modeling of process time series data based on the latest AI model

  • - Explore and apply AI methodologies such as RNN and transformer that are effective in sequence data analysis
  • - Exploration and design of model structures suitable for time series pattern modeling of process data
  • - Development of a sequence model that mimics the time series pattern of process data

- Sequence model analysis of process time series data based on XAI technique

  • - Exploring explainable AI methodologies for interpretation of sequence models
  • - Sequence model analysis of process time series data developed using XAI technique
  • - Analysis of the influence of features at each point in the process time series data

- Development of battery half-finished product characteristics and initial performance prediction model based on model analysis

  • - Model design suitable for battery semi-finished product characteristics and initial performance prediction model development
  • - apply the influence and correlation of the analyzed process data to the performance prediction model
  • - Development of battery semi-finished product characteristics and initial performance prediction model
대체 텍스트
Fig 1. Battery manufacturing process.
대체 텍스트
대체 텍스트
Fig 2. Interpretation of key parameters of mixed-type Mars capacity prediction model
대체 텍스트
Fig 3. t-SNE result of shipment capacity by model type.
대체 텍스트
Fig 4. Ensemble regression.
대체 텍스트
Fig 5. High performance of features based on voltage curves from the first 100 cycles.
대체 텍스트
Fig 6. Transformations of voltage–capacity discharge curves for three fast-charged cells that were tested with periodic slow diagnostic cycles.