In the production process of lithium-ion battery, the inconsistency of single cell characteristics in battery pack will seriously affect the whole life and performance of battery pack. In this paper, a lithium-ion battery grouping model is proposed, which includes a data preprocessing model and a novel battery grouping method of K-means algorithm to meet the specific requirements. In the battery data preprocessing model, the faulty battery is removed by designing the process data preprocessing method and actual production standards. In the battery matching algorithm, an improved K-means algorithm is proposed. After the battery grouping is completed, the number of batteries in each cluster is equal. Additionally, experiments for battery grouping and method validation are designed. Compared with the actual method, the proposed method proves its effectiveness.