IoT-Based Battery Health Monitoring

The problem

The battery is the most expensive component in an electric vehicle and it degrades unevenly. Fleet operators need to know which packs are ageing faster than expected before a cell failure strands a vehicle, and a periodic health check is too coarse to catch a trend developing.

How it works

Telemetry streams from vehicles over MQTT into AWS IoT Core, chosen because it holds up over the intermittent connectivity vehicles actually have rather than assuming a stable link. Time series models track degradation per pack against its own history, so a cell drifting from its expected curve is flagged even while its absolute readings still look acceptable.

What shaped it

Degradation is gradual and non-linear, which makes absolute thresholds a poor detector — by the time a reading looks bad in isolation, the trend has usually been visible for a long time. Modelling each pack against its own trajectory catches it earlier.

Fleet telemetry is also high volume and low value per message, so the pipeline was built to aggregate at the edge rather than ship every reading, keeping bandwidth and ingest costs proportionate to the insight.

Outcome

Degradation trends were predicted and at-risk cells flagged in real time, extending usable battery lifespan across the fleet.