A Hybrid Prognostic Approach for Remaining Useful Life Prediction of Lithium-Ion Batteries

Table 6

Comparison of the predicted RUL and sparse results for four lithium-ion batteries.

Battery name	Inspection cycle ()	Predicted RUL ()			Actual RUL ()	Relevant/support vectors
		SKE-RVM	SVM	BP		SKE-RVM	SVM	BP
A1	125	83	89	96	84	42	54	—
	146	63	66	71	63	46	62	—
	167	42	47	51	42	61	70	—
	188	20	25	27	21	60	75	—
A2	113	74	83	90	76	30	51	—
	132	58	61	66	57	40	55	—
	151	39	41	44	38	61	83	—
	170	20	25	27	19	52	86	—
A3	79	46	61	66	53	11	32	—
	92	38	45	48	40	23	38	—
	106	24	22	24	26	4	17	—
	119	13	15	16	13	4	15	—
A4	29	16	14	15	19	5	12	—
	34	14	16	18	14	4	14	—
	38	10	8	13	10	4	15	—
	43	5	6	7	5	6	18	—

SKE-RVM: selective kernel ensemble-based RVM learning algorithm.