Impedance estimation for lithium-ion battery using support vector regression and cuckoo search algorithm

With the expansion of the global new energy industry, lithium-ion batteries have been vigorously developed in several industries because of their high operating voltage, low cost, energy density, high power, long cycle life and excellent temperature properties. However, the performance of lithium-ion batteries will gradually undergo inevitable aging owing to usage time and operating environment. As an important parameter to recognize the health status of the battery, the impedance and its trend can measure the aging status of the battery. In this paper, incremental capacity (IC), Cuckoo Search Algorithm (CSA) and Support Vector Regression (SVR) are presented to assess the health status of the Li-ion batteries battery (Re and Rct). Firstly, it is clear that Re and Rct were tightly correlated with the health state of batteries, and the charging curve and IC curve of Li-ion batteries are analyzed, and the Savitzky-Golay approach is used to filter the IC curve. Then, five parameters were selected as the characteristic factors of Li-ion battery, and the correlation of the characteristic factors with Re and Rct was analyzed by Pearson correlation coefficient. Then, the mapping relationship between Re and Rct, the five feature factors was established based on SVR. The kernel parameters and penalty parameters of SVR were optimized by using CSA to improve the accuracy of the model. Finally, three lithium-ion battery aging experiments verify the presented method. The results reveal that the presented method can accurately estimate Re and Rct of different lithium-ion batteries, and its accuracy in estimating Re and Rct is the highest compared with other estimation methods.