Integrating mechanism and machine learning based capacity estimation for LiFePO4 batteries under slight overcharge cycling

Since the inconsistency in battery pack and failure of charging equipment, the slight overcharging for LiFePO4 batteries is occurred and even caused thermal runaway phenomenon. In this study, an integrated framework of aging mechanism and machine learning approach is utilized for capacity estimation of LiFePO4 batteries under slight overcharging cycling. Specifically, the non-destructive and post-mortem analyses are introduced to identify aging mechanism of LiFePO4 batteries. Meanwhile, the correlation between the internal chemical mechanism and incremental capacity curves is comprehensively conducted, and incremental capacity features are selected and verified as health indicators under slight overcharge cycling. The results show that the lithium dendrite and graphite-coated material separation accelerate the failure of LiFePO4 batteries under slight overcharge cycling. Moreover, the least square support vector machine is established for accurate capacity estimation of LiFePO4 batteries. When compared to the existing methods, the proposed approach can obtain higher accuracy in capacity estimation with a maximum relative error below 2%.