Hongzhang Xu

To ensure the safe and reliable application of lithium-ion batteries, accurate estimation of the state of health (SOH) is essential. However, most existing data-driven methods rely on charge/discharge data and lack interpretability due to their "black-box" nature. To address these challenges, this paper pro-poses an interpretable-algorithm-driven SOH estimation method based on relaxation voltage. To enhance the interpretability of the approach, we integrated the GradCAM++ algorithm into the con-volutional neural network (CNN). This integration not only im-proved the model’s estimation accuracy, with an average reduc-tion in root mean square error (RMSE) of 16.34%, but also gener-ated important class activation maps. Subsequently, these maps were correlated with a battery equivalent circuit model for aging analysis. The results reveal how the CNN focuses on either diffu-sion depolarization or activation depolarization across different battery lifecycle stages to perform SOH estimation. For the batter-ies exhibiting different aging trends, the model emphasizes distinct depolarization processes. Comprehensive validation experiments confirm that the proposed method achieves accurate SOH estima-tion with an average RMSE of 1.70% using only the first 6 minutes of relaxation voltage data, while providing insights for guiding SOH estimation strategies in practical applications.

State of health (SOH) estimation is critical for the safe and efficient operation of lithium-ion batteries. However, obtaining complete charge-discharge data in real-world applications remains challenging. This article proposes an SOH estimation framework based on physical information constraints and partial charging data reconstruction. A convolutional neural network (CNN) is designed to reconstruct partial voltage curves into the peak region of the incremental capacity (IC) curve, capturing critical aging information. Features extracted from the reconstructed IC curve are then used to estimate SOH, with physics-based constraints from reconstructed IC peak incorporated into the loss function to enhance model interpretability and accuracy. Experimental validation on three different datasets demonstrates that the proposed method achieves high estimation accuracy, with errors under 2% using limited voltage segments. The results verify the effectiveness and generalization capability of the approach, providing a practical solution for the limited data availability in real scenarios.

It is essential to accurately estimate the state of health (SOH) for lithium-ion batteries from the perspectives of safety and reliability. However, most existing data-driven methods are based on charging or discharging data, which is relatively difficult to apply. This paper proposes a novel SOH estimation approach based on the relaxation voltage reconstruction and a long short-term memory network with an attention mechanism (AM-LSTM) method. First, based on the relaxation voltage data, a complete relaxation curve is reconstructed with a gated recurrent unit (GRU) neural network. Next, health feature (HF) extraction is carried out on the reconstructed data, and the correlation is analyzed based on Pearson correlation analysis. Then, the SOH estimation is performed based on the AM-LSTM model. Various comparative studies have been conducted to verify the effectiveness of the proposed method by comparing it with relaxation voltage reconstruction and different SOH estimation methods. The experimental results demonstrate that the proposed method can effectively reconstruct the relaxation voltage and have good accuracy in estimating the SOH with a partial relaxation voltage curve.

To guarantee the safe and efficient operation of lithium-ion batteries, it is crucial to precisely estimate the state of health (SOH) of batteries. However, most of the existing studies have primarily focused on complete or large-range charging curves, which are highly challenging to acquire in practical applications. To this end, a novel SOH estimation method based on partial charging curve reconstruction is proposed in this article. First, a partial charging curve reconstruction model based on a convolutional neural network reconstructs the charging segments from voltage ranges with low SOH correlation to those with high SOH correlation. Second, health features are extracted from the reconstructed charging segments and used as inputs to estimate the SOH based on a Gaussian process regression model. Finally, validation experiments were conducted on two lithium-ion battery datasets to demonstrate the effectiveness and generalization of the proposed method. The proposed method enables precise SOH estimation using only a narrow charging segment (0.05 V), making it suitable for practical application scenarios.

It is vital to accurately estimate the state of health (SOH) of lithium-ion batteries of electrical vehicles. Despite the significant impact of temperature on the SOH, most conventional data-driven methods for SOH estimation neglect the influence of temperature-related characteristics. This paper proposes a novel SOH estimation method with attentional feature fusion (AFF) considering differential temperature (DT) features for lithium-ion batteries. Firstly, this paper extracts DT features from the differential temperature curves and employs the Pearson correlation coefficient (PCC) to select the health features (HFs) of the DT curve that exhibit the strongest correlation with SOH. Then, a novel SOH estimation model combining AFF and LSTM (AFF-LSTM) is proposed, which can automatically learn the influence weights of different types of HFs on SOH, thereby improving the accuracy of SOH estimation. The MAPE evaluation metric is approximately 0.126% on the Oxford dataset. Finally, various comparative analyses demonstrate that the proposed method can accurately estimate the SOH of lithium-ion under different operating temperatures.