Abstract
State of health estimation of lithium-ion battery based on data-driven methods are influenced by the model input. Moreover, utilizing multi-dimensional health indicators as model input to estimate battery state of health does not increase the estimated accuracy, while it will increase the computational burden of the battery model. Therefore, this paper presents a prognostic framework based on principal component analysis technique to decrease the number of model input, while the health indicators are extracted from battery incremental capacity curves. Besides, the relationship between health indicators and state of health is established by Gaussian process regression. In addition, considering the diversity of the battery operating condition, this paper analyses the prediction of battery state of health under the three aging temperature mode, which consider four verified types. Moreover, using eight lithium-ion cells which aged at three different conditions to test the performance of our approach. The proposed methods have a higher accuracy based on the results of three types of errors, which show it can get less than 2.5% estimated error.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have