Impact of Data Partitioning to Improve Prediction Accuracy for Remaining Useful Life of Li-Ion Batteries

Table 1

Prediction accuracies with each data partitioning method when the percentage of the training set is varied.

Cell	Percentage; (%)	INI			BOX			ADD
		MAE (cycle)	RMSE (cycle)	MAPE (%)	MAE (cycle)	RMSE (cycle)	MAPE (%)	MAE (cycle)	RMSE (cycle)	MAPE (%)
LFP	20	104	125	0.581	104	125	0.587	103	122	0.546
	30	61	77	0.402	64	77	0.385	61	75	0.381
	40	44	53	0.290	62	72	0.336	41	49	0.235
	50	55	65	0.452	136	146	0.830	51	60	0.390
	60	49	60	0.582	94	106	0.932	46	57	0.552
	70	57	65	0.744	77	85	0.800	58	66	0.744
	80	44	49	0.693	45	53	0.596	44	49	0.692
	90	32	33	0.711	22	24	0.463	30	32	0.671
NCA	20	87	103	2.845	87	103	2.849	87	103	2.849
	30	76	89	2.658	87	101	3.076	76	89	2.666
	40	65	74	2.457	76	87	2.967	65	75	2.467
	50	58	64	2.354	69	76	2.902	58	65	2.376
	60	31	35	1.670	41	47	2.259	31	35	1.673
	70	29	32	1.763	40	44	2.417	29	32	1.755
	80	19	20	1.448	26	28	2.042	19	20	1.451
	90	9	9	0.981	15	16	1.703	9	9	0.971
NMC	20	100	115	3.029	100	115	3.155	105	121	3.132
	30	60	68	2.213	68	79	2.617	61	70	2.271
	40	43	48	1.801	57	66	2.529	44	50	1.844
	50	30	34	1.538	44	50	2.239	32	35	1.572
	60	24	26	1.409	38	43	2.208	24	26	1.424
	70	20	21	1.420	36	39	2.451	20	21	1.428
	80	16	17	1.539	26	27	2.451	16	17	1.543
	90	13	13	1.889	25	27	2.320	13	13	1.906