Battery cell consistency evaluation method based on charging fragment data and unsupervised algorithm

The invention discloses an unsupervised standard battery cell consistency fault detection method based on charging fragment data, and the method comprises the following steps: S1, data preparation and exploration: exploring and analyzing the data distribution condition of the maximum voltage difference in a single battery cell for three types of batteries, and carrying out the subsequent practice of a data-driven algorithm according to the actual data distribution of the three types of batteries; s2, feature engineering: preprocessing the data, screening effective features through a feature extraction and principal component analysis dimensionality reduction method, and performing subsequent modeling, including data preprocessing, feature extraction and PCA dimensionality reduction; and S3, model construction: adopting a plurality of unsupervised learning algorithms, carrying out combined modeling aiming at the charging and discharging characteristics, and comparing the validity of different algorithms. Accor