BatteryGuard AI is a predictive tool that uses AI to monitor the health and safety of second-life lithium-ion batteries. It helps users—especially in community energy projects—predict battery failures, estimate remaining life, and prevent safety risks. The system provides clear, explainable outputs so non-experts can understand and trust the insights.

Inspiration

This project was inspired by the growing use of second-life batteries in off-grid and low-resource areas. While these reused batteries are sustainable and affordable, they can fail without proper monitoring. BatteryGuard AI empowers users with a safe, smart, and responsible way to manage energy storage using AI.

What it does

BatteryGuard AI predicts the health and remaining life of second-life lithium-ion batteries using historical data (voltage, temperature, current). It detects early warning signs of failure, flags risks, and explains its decisions—making battery safety understandable for non-experts.

How we built it

We trained an Neural Network model using publicly available datasets (e.g., NASA, CALCE) to forecast battery degradation and detect anomalies. A Streamlit-based dashboard visualizes predictions, health scores, and interpretable alerts powered by SHAP. AWS SageMaker was used for model deployment.

Challenges we ran into

Finding high-quality open data for second-life batteries Balancing model accuracy with explainability Designing a simple interface for users without technical backgrounds

Accomplishments that we're proud of

Building a working prototype with real-time predictions Integrating explainable AI for battery health decisions Aligning our solution with responsible AI principles

What we learned

How to apply time-series AI to real-world battery problems The importance of interpretability in safety-critical systems How cross-disciplinary collaboration strengthens social impact solutions

What's next for BatteryGuard AI

Expand training on real second-life battery data Collaborate with NGOs and clean energy startups for pilot testing Package the tool as an open-source solution or low-cost SaaS for the energy access market

Built With

Share this project:

Updates