Lithium-ion batteries (LIB) are considered the leading technology for storage systems that can be applied to fast vehicles and stationary systems such as solar panels. However, the technology has higher density and capacity than other technologies; it must be operated under controlled parameters of voltage, current, and temperature to avoid the appearance of abuses and failures. Therefore, diagnostic and prognostic models based on data-driven (DD) have been applied to predict and forecast the possible failures that can appear in the cells. Thus, LIB is a multidisciplinary problem that involves electrical, mechanical, and computer engineering. Despite the extensive adoption of ML models, this approach is sensible to over-fitting and bias that can reduce the accuracy of the models. One solution is Ensemble Learning (EL), which combines several single learners to obtain a more robust response. Therefore, this work applied several EL models to classify failures in LIB. The EL models were involved in two datasets, and the results were compared with single models. Every EL performs better on average than the single models, especially the Bagging, RF, Boosting, and Stacking approaches. Finally, the results proved that EL improves the response accuracy for LIB problems.