The grouping and large-scale of battery energy storage systems lead to the problem of inconsistency. Practi-cal consistency evaluation is significant for the management, equalization and maintenance of the battery system. Various evaluation methods have been developed over the past decades to better assess battery pack consistency. In these research efforts, the accuracy of the assessment results is often of paramount importance. In this work, a battery pack consistency evaluation approach is proposed based on multi-feature information fusion. Ohmic resistance, polarization resistance and open circuit volt-age are identified as feature parameters from electric vehicle operation data. An adaptive forgetting factor recursive least squares (AFFRLS) algorithm is developed using fuzzy logic to modify the forgetting factor for parameter identification. Grey correlation analysis is applied to calculate the dispersion of features (DF). The DF is weighted to evaluate the inconsistency of the battery pack. Further, the weights are assigned through the CRITIC-G1 method. Moreover, a mapping model between the extracted voltage features and the DF is established through a cost-sensitive support vector machine (CS-SVM) algorithm, which is used to evaluate and predict the consistency distribution of battery parameters. Finally, the proposed algorithm is verified by experimental data. The results indicate that the proposed parameter identification, consistency evaluation and prediction methods have high accuracy. This work was supported in part by the National Natural Science Foundation of China under Grant 62203352, U2003110, U2106218, 52107205, and in part by the Key Laboratory Project of Shaanxi Provincial Department of Education (No. 20JS110).