杨博论文“A novel state of health estimation method for lithium-ion battery based on partial incremental capacity and Support vector regression”发表在外文期刊《IOP Conference Series: Earth and Environmental Science》上

State of Health (SOH) is critical for lithium-ion batteries as it ensures the safety of batteries’ health condition and provides a basis for retirement of the batteries. In order to provide an accurate estimation of the SOH, a novel hybrid estimation method based on the partial incremental capacity and Support vector regression (SVR) is proposed in this paper. Firstly, the Savitzky-Golay method is applied to smooth the initial incremental capacity curves under the period of constant current charge. Then the key health features are extracted from the partial incremental curve theoretically and selected through correlation analysis methods. Finally, an SVR model is constructed to estimate the SOH. Several battery datasets under different cycling test conditions are used to validate the effectiveness of the proposed method. The result shows that the proposed method can provide a reliable and accurate estimation for SOH.

The IC curve can be obtained in each charge/discharge cycle with voltage ranging from 2.5V to 3.65V while charging. In practice, we use the partial incremental capacity curve of which voltage ranges from 3.25V to 3.45V to extract features as shown in Fig.2. 15 features can be extracted from the partial IC curve totally.

The SVR is a machine learning algorithm which is efficient in nonlinear modeling and time series forecasting[6]. The goal of the algorithm is to find the mapping relation between input data and output data under the assumption that the joint distribution of the input data and output data is unknown. The kernel-trick is used in the algorithm to transform the complex nonlinear problem into a simple linear problem.

To verify the superiority of the proposed algorithm, three other machine learning algorithms, Xgboost, Decision tree and Adaboost are employed to compare with the SVR. The comparison results are shown in Table.3. The results show that the proposed approach can achieve higher accuracy and consistency than other algorithms.

该研究在金老师、马晓明老师共同指导下，由杨博同学完成，成果发表在外文期刊《IOP Conference Series: Earth and Environmental Science》上。

文章链接：https://doi.org/10.1088/1755-1315/804/4/042004