- Tytuł:
- Similarity Based Remaining Useful Life Prediction for Lithium-ion Battery under Small Sample Situation Based on Data Augmentation
- Autorzy:
-
Wang, Zongyao
Shangguan, Wei
Peng, Cong
Cai, Baigen - Tematy:
-
lithium-ion batteries
remaining useful life
similarity-based prediction
Sobol sampling
kernel density estimation - Pokaż więcej
- Wydawca:
- Polska Akademia Nauk. Polskie Naukowo-Techniczne Towarzystwo Eksploatacyjne PAN
- Powiązania:
- https://bibliotekanauki.pl/articles/59112846.pdf Link otwiera się w nowym oknie
- Opis:
- Lithium-ion batteries find extensive application in transportation, energy storage, and various other fields. However, gathering a significant volume of degradation data for the same type of lithium-ion battery devices becomes challenging in practice due to variations in battery operating conditions and electrochemical properties, among other factors. In this small sample situation, accurately predicting the remaining useful life (RUL) of the battery holds great significance. This paper presents a RUL prediction method that is based on data augmentation and similarity measures. Firstly, by utilizing the single exponential model and Sobol sampling techniques, it is possible to generate realistic degradation trajectories, even with just one complete run-to-failure degradation dataset. Subsequently, the similarity between the generated prediction reference trajectories and actual degradation trajectories is evaluated using the Pearson distance. Following that, the point estimation of RUL is performed through weighted averaging. Then, the uncertainty of the RUL predictions is quantified using kernel density estimation. Finally, the effectiveness of the proposed RUL prediction method is validated using two NASA lithium-ion battery datasets. Results demonstrate the practicality and effectiveness of the proposed method.
- Dostawca treści:
- Biblioteka Nauki
Artykuł