AI-Driven Solid-State Battery Discovery & Health Prediction
Accelerate Battery Innovation with AI-Driven Materials Discovery and Predictive Modeling
Skills you will gain:
About Program:
This three-day hands-on workshop introduces AI-driven approaches for solid-state battery materials discovery and battery health prediction. Participants will work with research-grade materials databases and real battery cycling datasets to explore materials data mining, machine learning, physics-informed modeling, and battery degradation analysis.
Aim: To equip participants with practical AI and machine-learning workflows for solid-state battery materials discovery, electrolyte-property prediction, physics-informed battery modeling, and SoH/RUL prediction using real scientific datasets.
Program Objectives:
- Understand solid-state battery technologies and solid electrolyte properties.
- Retrieve and curate battery-material data from Materials Project and OQMD.
- Generate materials descriptors using Pymatgen and Matminer.
- Develop machine-learning models for electrolyte-property prediction.
- Apply XGBoost, Random Forest and SHAP for prediction and explainability.
- Understand foundational Physics-Informed Neural Network (PINN) workflows.
- Analyze battery cycling data and degradation indicators.
- Develop SoH and RUL prediction workflows using time-series and LSTM modeling.
What you will learn?
Day 1: Solid-State Battery Fundamentals & Materials Data Preparation
- Fundamentals of lithium-ion and solid-state batteries
- Solid electrolytes, ionic conductivity, stability, and key performance factors
- Introduction to AI and materials informatics in battery research
- Materials Project and OQMD for battery-material data
- Material structure processing using Pymatgen
- Feature and descriptor generation using Matminer
- Building an AI-ready solid-electrolyte dataset
- Basic visualization and screening of candidate materials
Hands-On: Create and explore an AI-ready solid-electrolyte materials dataset.
Day 2: Machine Learning & Explainable AI for Materials Discovery
- Data preprocessing and feature selection
- Ionic-conductivity and materials-property prediction
- Random Forest and XGBoost modeling
- Model validation using MAE, RMSE, R², and cross-validation
- Predicted vs actual performance analysis
- Explainable AI using SHAP
- Identification of important material descriptors
- AI-based ranking of promising solid-electrolyte candidates
- Introduction to physics-informed machine learning for battery research
Hands-On: Build, evaluate, and interpret an explainable AI model for solid-electrolyte screening.
Day 3: Battery Health, SOH & Remaining Useful Life Prediction
- Introduction to battery cycling and degradation datasets
- Capacity fade and battery-aging analysis
- Feature engineering from voltage, current, capacity, temperature, and cycle data
- State of Health (SOH) estimation and prediction
- Baseline ML models for battery-health prediction
- Time-series modeling using LSTM
- Battery degradation forecasting
- Remaining Useful Life (RUL) estimation
- Comparison of machine-learning and deep-learning models
- Research interpretation and identification of future experimental directions
Hands-On: Develop an AI workflow for SOH prediction, degradation forecasting, and RUL estimation.
Mentor Profile
Fee Plan
Get an e-Certificate of Participation!
Intended For :
- MSc/M.Tech students in materials science, chemistry, physics, chemical engineering, energy engineering, or related fields
- PhD scholars and postdoctoral researchers in battery, materials, energy-storage, and computational research
- Faculty members, academicians, and research scientists
- Computational materials and machine-learning researchers
- Battery, EV, energy-storage, and BMS professionals
- Industry R&D professionals working in battery technologies and advanced energy systems
Career Supporting Skills
Program Outcomes
- Extract and curate solid-state electrolyte data from research-grade materials databases.
- Generate materials descriptors and features using Pymatgen and Matminer.
- Develop and validate machine-learning models for electrolyte-property prediction.
- Interpret ML predictions using feature importance and SHAP-based explainable AI.
- Apply foundational PINN workflows to physics-informed battery modeling.
- Process and analyze battery cycling data to identify degradation and health indicators.
- Develop SoH and RUL prediction models using time-series and LSTM-based approaches.
- Build reproducible AI workflows in Google Colab for battery and materials research.
- Interpret model outputs in the context of battery materials, energy storage, and EV applications.