Identification of the Current State of a Battery using Impedance Measurements

The role of electric vehicles (EVs) in public and private transportation is being redefined. Motivated by the current and looming climate crisis, EVs are seen as a bridge technology to more energy-efficient modes of transportation. A key aspect of widespread adoption is meeting the expectations of users who are accustomed to the comfort, safety and range of a gasoline-powered car. To achieve this, an accurate understanding of the current state of the battery is essential. As battery science has progressed, better cell types have been developed, each with its own unique characteristics and challenges. While lithium ion battery cells are very much discussed as the best available technology for energy storage in EVs, some conventional state of charge (SoC) detection methods are not applicable. This thesis investigates the use of impedance-based SoC estimation in EVs. This work focuses on battery state detection based on a cell model and test drive measurements. A method is proposed that aims to overcome the shortcomings of the traditionally used Fourier transform (FT) with respect to the effects caused by signals occurring simultaneously with the SoC measurements in the on-board power supply network. While the performance in this particular aspect does not exceed that of the FT, the low losses on gappy time series signals prove to be an outstanding advantage of the proposed method. The investigations are based on artificial and measured data. As an SoC classification method, a Naive Bayes classifier is applied, using the output of the proposed algorithm as input features.