Prognostics for state of health estimation of lithium-ion batteries based on combination Gaussian process functional regression
Citations
1,116 citations
Cites background from "Prognostics for state of health est..."
...[351] improved long-term prediction performance of GPR by combining two covariance functions to capture the actual trends of both global degradation and local regeneration....
[...]
538 citations
Cites methods from "Prognostics for state of health est..."
...[141] utilized a combination of covariance functions and mean functions in GPR for multi-step-ahead prognostics....
[...]
471 citations
399 citations
Cites background or methods from "Prognostics for state of health est..."
...Batteries 5, 6, 7, and 18 (in the numbering of the online repository) were chosen to be analysed in the present work, since these have the most data-points; and because they have previously been chosen for analysis in earlier works [8,10], and hence the present selection facilitates a comparison with those works....
[...]
...[10] and used with the explicit mean function GP...
[...]
...[10] applied Gaussian process regression to battery capacity prediction, and showed that their predictive accuracy was improved when a linear or quadratic Explicit Mean Function (EMF; see Section 2....
[...]
...In particular, it is favourable over previous GP capacity estimation methods that use data from identical cells, which merely identify an optimal prior estimate for the parameters of a parametric model, which are then updated sequentially [10]....
[...]
...Kernel function selection is perhaps the most important aspect of GPmodelling, yet it has not been addressed in a principled manner in the aforementioned battery degradation literature [6,10,15]....
[...]
393 citations
References
11,357 citations
"Prognostics for state of health est..." refers background or methods in this paper
...It can model the behavior of any system through the combination of the appropriate Gaussian process and realize prognostics combined with prior knowledge based on a Bayesian framework [28]....
[...]
...A periodic covariance function is generally used to model a function within a specific period [28]....
[...]
...Here we list the covariance functions that we applied in battery health prognostics [28,32]....
[...]
..., n for the training means [28] and for the test means u⁄....
[...]
...Generally, the hyper-parameters are needed to be optimized with the maximization of the log-likelihood function given by [28,32]:...
[...]
1,225 citations
795 citations
692 citations
601 citations