A state of health estimation method of lithium-ion batteries based on DT-IC-V health features extracted from partial charging segment

To ensure a safe and reliable operation, accurate estimation of the state of health (SOH) of lithium-ion batteries is necessary. In order to improve the accuracy and practicability of the SOH estimation, this paper proposes a SOH estimation method based on differential temperature-incremental capacity-voltage (DT-IC-V) health features (HFs). A new DT-related health feature extraction method is proposed by analyzing the potential relationship between the temperature difference profile and SOH. A set of DT-IC-V HFs are designed in a relatively small charging segment to reduce the difficulty of obtaining data in practice. And a battery SOH estimation model based on deep belief network (DBN) and extreme learning machine (ELM) is designed. The number of nodes in each hidden layer of the DBN-ELM model is determined by the Sparrow Search Algorithm (SSA). The proposed method is validated on different types of batteries. The results show that the method can accurately estimate the SOH, with the mean absolute percent error remaining within 0.43% and 1.35% in the Oxford and NASA datasets, respectively.