In this paper, battery aging diversity among independent cells was studied in terms of available capacity degradation. During the aging process of LiFePO4 batteries, the phenomenon of aging diversity can be observed. When batteries with same specification were charged and discharged repeatedly under the same working conditions, the available capacity of different cell decreased at different rates along the cycle number. In this study, accelerated aging tests were carried out on multiple new LiFePO4 battery samples of different brands. Experimental results show that under the same working conditions, the actual available capacity of all cells decreased as the number of aging cycle increased, but an obvious aging diversity was observed even among different cells of same brand with same specification. This aging diversity was described and analysed in detail, and the common aging features of different cells beneath this aging diversity was explored. Considering this aging diversity, a probability density concept was adopted to estimate battery \' s state of health ( SOH) . With this method, a relationship between battery SOH and its aging feature parameter was established, and a dynamic sliding window optimization technique was designed to ensure the optimal quality of aging feature extraction. Finally, the accuracy of this SOH estimation method was verified by random test.