Tackling Limited Labeled Field Data Challenges for State of Health Estimation of Lithium-Ion Batteries by Advanced Semi-Supervised Regression
Bibliographic record
Abstract
Accurate estimation of battery state of health (SOH) has become indispensable in ensuring the predictive maintenance and safety of electric vehicles (EVs). While supervised machine learning excels in laboratory settings with adequate SOH labels, field-based SOH data collection for supervised learning is hindered by EVs' complex conditions and prohibitive data collection costs. To overcome this challenge, a battery SOH estimation method based on semi-supervised regression is proposed and validated using field data in this paper. Initially, the Ampere integral formula is employed to calculate SOH labels from charging data, and the error of labeled SOH is reduced by the open-circuit voltage correction strategy. The calculation error of the SOH label is confirmed to be less than 1.2%, as validated by the full-charge test of the battery packs. Subsequently, statistical features are extracted from charging data, and health indicator sets are selected by two correlation analysis methods (Pearson correlation and grayscale correlation). Moreover, two regressors are trained by learning the mapping between labeled SOH and various health indicator sets. To enhance the training dataset, semi-supervised with co-training is utilized to estimate pseudo-labels for unlabeled charging data. The final SOH estimation is achieved through the fusion of these two regressors. Finally, the proposed method is validated using field data from 20 electric forklifts collected over approximately one year. Remarkably, even with only 10 labeled data points, the proposed method achieves a mean absolute error in SOH estimation of a mere 3.96%. This represents a significant reduction of 20% compared to the traditional supervised learning method. Compared with the two benchmarks without co-training, the estimation error drops by 7.69% and 8.76%, respectively.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".