Battery Thermal Digital Twin
Estimating internal temperature from noisy measurements
Key idea
Battery internal temperature cannot be measured directly.
But it can be estimated.
System
Four parts:
- Thermal model → heat generation + cooling
- Sensor → noisy temperature
- Estimator → reconstructs true temperature
- Metrics → efficiency and degradation
What I did
- Simulated heating from current (I²R)
- Added sensor noise
- Built a recursive estimator
- Compared two degradation models:
- threshold
- exponential
- threshold
What you see
- True temperature (hidden state)
- Noisy sensor signal
- Estimated temperature (digital twin)
Key results
- Estimator recovers smooth temperature from noisy data
- Higher current increases temperature and degradation
- Model choice changes degradation behavior
Why it matters
Real systems give noisy data.
You still need the true state.
Media
Code
- GitHub: https://github.com/YOUR_USERNAME/battery-digital-twin