State of health estimation for lithium-ion batteries based on recurrence plot analysis using charging voltage curves

Accurate state of health (SOH) estimation is crucial for ensuring the reliability and safety of lithium-ion batteries (LIBs) in various applications. Traditional SOH estimators often require complex models, strict test profiles (e.g., low cycling rates), or extensive training data, limiting their applications. Hence, this study introduces a novel data-driven method that integrates recurrence plot (RP) analysis for direct SOH estimation using a simple regression model trained on a single battery. The method begins by transforming voltage data from charging cycles into RP images, which serve as a basis for extracting significant features associated with LIB degradation. These features, termed features-of-interest (FOIs), are then thoroughly analyzed to establish their correlation with the aging status of LIBs. Experimental results demonstrate that the proposed RP-derived FOIs align closely with those obtained from differential voltage analysis (DVA) using open circuit voltage (OCV) data, even under high charging rates. This alignment underscores the effectiveness of the FOIs as reliable indicators of SOH. Moreover, SOH estimations conducted on battery aging datasets reveals that RP-derived FOIs from a single LIB can accurately estimate the SOH of all remaining LIBs through a simple regression model, reducing the need for extensive training data and computational resources.