Skip to main content
EV Evidence

Maintenance · UK

Battery evidence method

Battery claims come only from the current E2 battery evidence artefact and retain the study's population, metric, conditions and limitations. A cohort estimate or model error is not a measured State of Health for an individual vehicle. The admitted evidence does not support a universal degradation rate, winter-loss percentage, lifetime forecast or model-specific interpolation.

IEEE Access: Bayesian Network Based State-of-Health Estimation for Battery on Electric Vehicle Application and its Validation Through Real-World Data

Reviewed

Key facts and evidence

For UK battery content using the E2 evidence snapshot of 12 August 2026, the only admitted numerical claim is an estimator result from 16 same-model, high-use taxis: mean absolute error below 4%. That is model validation, not measured degradation. The evidence supplies no universal fade or winter-range percentage, and it cannot predict an individual vehicle without vehicle-specific assessment.

IEEE Access: Bayesian Network Based State-of-Health Estimation for Battery on Electric Vehicle Application and its Validation Through Real-World Data

How EV Evidence admits battery studies, separates metrics and labels cohort estimates without presenting them as measured vehicle State of Health.

Evidence admission

Numerical battery claims are drawn only from the current E2 battery evidence artefact. A study must have an explicit reuse basis and an eligible claim record. Excluded studies may remain in audit counts but do not supply page numbers, charts or inferred averages.

Metric separation

Usable-capacity State of Health, laboratory relative capacity, resistance growth, dashboard range, modelled range retention, estimator error and replacement occurrence are different metrics. Their values and denominators are never substituted for one another. Estimate and measurement labels remain visible in summaries and exports.

Admitted estimator claim

Huo and colleagues trained and validated an estimator within 16 BAIC EU260 taxis followed for two years. E2 admits the reported mean absolute error below 4%, with its same-fleet and high-use limitations. EV Evidence does not rewrite that error as battery degradation, range loss or transfer performance for another model.

Limits and individual use

No cross-study average, annual fade rate, winter-range loss, remaining lifetime or model interpolation is produced. Unknown bounds remain unknown rather than zero. A vehicle-specific measured result must identify the vehicle, date, conditions, tool or method, metric and limitations before it can inform an individual decision.

What to do next

Use this method to judge a published battery claim, not to diagnose a pack or infer that a vehicle is fit for use.

Immediate action: Request a vehicle-specific written assessment when battery condition materially affects a purchase or repair decision.

  • Stop relying on an estimate if the vehicle identity, test method, date, conditions, metric or limitations are missing.

Professional hand-off

  • Use an independent EV-competent battery assessor or manufacturer-authorised specialist for a vehicle-specific report and explanation.

Questions this record can answer

What does the sourced evidence say about evidence admission?

Numerical battery claims are drawn only from the current E2 battery evidence artefact. A study must have an explicit reuse basis and an eligible claim record. Excluded studies may remain in audit counts but do not supply page numbers, charts or inferred averages.

What does the sourced evidence say about metric separation?

Usable-capacity State of Health, laboratory relative capacity, resistance growth, dashboard range, modelled range retention, estimator error and replacement occurrence are different metrics. Their values and denominators are never substituted for one another. Estimate and measurement labels remain visible in summaries and exports.

What does the sourced evidence say about admitted estimator claim?

Huo and colleagues trained and validated an estimator within 16 BAIC EU260 taxis followed for two years. E2 admits the reported mean absolute error below 4%, with its same-fleet and high-use limitations. EV Evidence does not rewrite that error as battery degradation, range loss or transfer performance for another model.

Sources