A hybrid intelligent model using the distribution of relaxation time analysis of electrochemical impedance spectroscopy for lithium-ion battery state of health estimation

The state of health (SOH) estimation of lithium-ion batteries is essential to ensure the safety of electric vehicles. Electrochemical impedance spectroscopy (EIS) measurement can provide valuable ageing information and pave the way for the battery SOH estimation. However, the semicircle's overlapping in the EIS during battery degradation introduces large uncertainty of the electrochemical process, causing difficulty in identifying battery ageing features. Therefore, this paper proposes a hybrid intelligent model for the extraction of highly influential health indicators (HIs) and more accurate lithium-ion battery SOH estimation. The distribution of relaxation time is first used to interpret the EIS and distinguish various electrochemical processes at multi-time scales. Second, various physical HIs in different ageing processes are further extracted based on the autoencoder and Spearman rank correlation analysis. Third, an improved cascade feedforward neural network using the genetic algorithm (GA-CF) is proposed to develop an accurate SOH estimation model using the extracted HIs. Finally, the developed model is validated at different operating temperatures of the lithium-ion battery, including 25(degrees)C, 35 C-degrees, and 45(degrees)C. The results reveal that the developed GA-CF model with the proposed physical HIs can achieve more accurate and robust SOH estimation, compared to raw features or other popular models.