New framework extracts health features from BESS field fault data
Capacity, degradation rate, and dV/dQ peaks beat resistance for fault detection.
Deep Dive
Researchers analyzed operational data from 25 commercial grid-connected lithium-ion BESS modules (14 series-connected parallel groups each, 25 faulty, 325 non-faulty). Their framework extracts calibrated health features including group-level capacity, capacity degradation rate, and dV/dQ peak heights, which separate faulty parallel-connected cell groups with statistical significance (p<0.05). Internal resistance did not differentiate faults (p>0.05), challenging exclusive reliance on resistance for fault detection.
Key Points
- Analyzed 25 commercial BESS modules (14 parallel groups each) with 25 confirmed faulty and 325 non-faulty cell groups
- Capacity, capacity degradation rate, and dV/dQ peak heights statistically separate faults (p<0.05)
- Internal resistance did not differentiate faults, challenging conventional resistance-based detection methods
Why It Matters
Improves early fault detection in grid-scale batteries, preventing catastrophic failures and reducing reliance on unreliable resistance metrics.