Real-time battery health monitoring dashboard for electric vehicle fleets. Monitors forklifts, delivery vans, cargo bikes, and e-scooters.
Built by Prateek Gaur — Forward Deployment Engineer | Battery Systems | AI Portfolio: https://prateek-gaur-ml-bz0s69q.gamma.site GitHub: https://github.com/PRATdoppelEK
A production-ready battery fleet intelligence system that gives fleet operators instant visibility into the health of every battery in their fleet.
No more discovering a failed forklift battery at the start of a shift. No more unexpected delivery van breakdowns mid-route. No more replacing batteries too early or too late.
The dashboard shows:
- Live fleet status overview (Critical / Warning / Healthy counts)
- Active alerts with recommended action for each vehicle
- SOH distribution per vehicle type with EOL threshold line
- SOC vs SOH scatter plot to identify battery patterns
- Temperature distribution with thermal risk zones
- Full fleet table with sortable columns and maintenance actions
Step 1 — Install dependencies
pip3 install streamlit plotly pandas numpy scikit-learn --break-system-packages
Step 2 — Run the dashboard
python3 run.py
Step 3 — Open browser at http://localhost:8501
For terminal-only mode (no browser needed):
python3 run.py --terminal
| Type | Count | Battery | Cycles/Day | Typical Use |
|---|---|---|---|---|
| Electric Forklift | 8 | 48V 500Ah | 2 | Warehouse operations |
| Delivery Van | 6 | 400V 200Ah | 1 | Last-mile delivery |
| Cargo Bike | 10 | 48V 15Ah | 3 | Urban delivery |
| E-Scooter | 16 | 48V 7Ah | 4 | Micro-mobility |
Total fleet: 40 vehicles
| Status | Meaning | Action |
|---|---|---|
| CRITICAL | Thermal runaway risk or severe fault | Remove from service immediately |
| WARNING | Cell degradation or resistance fault | Inspect within 48 hours |
| REPLACE_SOON | SOH below 80% threshold | Plan replacement within 30 days |
| LOW_BATTERY | SOC below 15% | Return to charging station |
| HEALTHY | Operating within normal parameters | No action required |
SOH (State of Health): ratio of current capacity to rated capacity. Standard end-of-life threshold: SOH = 80%. Estimation method: capacity fade model based on total cycles and degradation rate.
SOC (State of Charge): current energy level 0-100%. Estimation method: Coulomb counting with temperature-corrected OCV.
Anomaly detection: statistical deviation from fleet average per vehicle type.
This project builds on the author's M.Sc. thesis (Custom Battery Cell Balancing Under Thermal Gradient, TU Berlin) and industrial experience at Dan-Tech Energy GmbH (LSTM-based SOH prediction on real industrial datasets).
ev-fleet-battery-intelligence/
|-- data/
| |-- simulator.py # Realistic fleet data generator
|-- models/
| |-- health_analyzer.py # SOH analysis, RUL estimation, alert logic
|-- dashboard/
| |-- app.py # Streamlit web dashboard
|-- reports/
| |-- report_generator.py # Automated fleet health report
|-- docs/
| |-- fde-deployment-guide.md # How to deploy at a real customer
|-- run.py # Single entry point
|-- requirements.txt
- battery-soh-lstm: LSTM model for SOH prediction (MAE=0.018, R2=0.94)
- battery-ecm-simulation: Python ECM with 14S3P pack simulation
- battery-digital-twin: Full MLOps pipeline with Kubernetes and drift monitoring
- enterprise-industrial-agent-harness: FDE toolkit for enterprise deployments
Prateek Gaur — Forward Deployment Engineer | Battery Systems | AI Enablement
📧 prateekgaur@gmx.de
🌐 Portfolio
💼 LinkedIn
🐙 GitHub
See also:
battery-soh-lstm ·
battery-ecm-simulation ·
battery-digital-twin ·
enterprise-industrial-agent-harness