State of Health Estimation for Lithium-Ion Battery Using Empirical Degradation and Error Compensation Models

Abstract

State of health (SOH) estimation is always an important factor in ensuring the reliability and safety of lithium-ion batteries. In view of the shortcomings of the existing SOH estimation methods, such as non-universal, the estimation of different batteries is limited, and the accuracy is insufficient. A fusion estimation method that depends on an empirical degradation model and a data-driven method is proposed. First, we construct an empirical degradation model of lithium-ion battery SOH with charge-discharge cycles. Four working condition characteristics are extracted from the actual charging and discharging process of batteries. Then, with these features as inputs, the prediction error of the empirical degradation model is taken as the output, and training the error compensation model becomes dependent on the data-driven method. The actual working condition characteristics of the tested lithium ion battery are substituted into the training error compensation model, and the model output is fed back to the prediction results of the empirical degradation model. A high-precision estimation of lithium-ion battery SOH is thereby achieved. Finally, the proposed method is verified based on the NASA lithium-ion battery data set. The results show that the fusion method is applicable to different lithium-ion batteries of the same type, and the mean absolute percentage error of SOH estimation is approximately 2%, indicating that the proposed method exhibits good estimation performance and applicability.